Sources

console.typesafe.ai

TypeSafe drops the Jev waitlist; new accounts start with $5

console.typesafe.ai

TypeSafe starts pulling people off the Jev waitlist

docs.typesafe.ai

TypeLLM said Jev 1.14 cannot return every vehicle box, including Tesla's flying Roadster; this desk's every-vehicle Box was one rectangle, n = 3, and type array returned HTTP 400

docs.typesafe.ai

What Jev returns: System One in plain language

docs.typesafe.ai

TypeLLM said Jev 1.14 cannot return every vehicle box, including Tesla's flying Roadster; this desk's every-vehicle Box was one rectangle, n = 3, and type array returned HTTP 400 A green Continue button scored noul 0.99 on jev-1.14.0, n = 3, and an empty field scored 0.02

docs.typesafe.ai

TypeSafe lists nine ways Jev 1.13 is still jagged, and asks for more

docs.typesafe.ai

TypeSafe-Danger-Zone-Long-Context raises both Jev context caps to 262k tokens

docs.typesafe.ai

On DeepEval's Paris weather example, jev-1.13.0 scored 0.254 when the answer invented a breeze On 40 client requests, Jev and Opus both scored 37 of 40, at $0.0015 against $0.005 to $0.015

docs.typesafe.ai

On eight Stripe passages, a 0-9 Score put Roll secret at 8.86 and the other six at 2.42 or below

docs.typesafe.ai

One builder ran Jev on 1,500 of his own emails

docs.typesafe.ai

What Jev returns: System One in plain language

docs.typesafe.ai

On October 26 TypeSafe's jev-latest named jev-1.14.0, and both aliases point at 1.14.0 On WANDS, more Jev questions move the mean to 0.610 and leave the median at 0.561

docs.typesafe.ai

TypeSafe drops the Jev waitlist; new accounts start with $5 OpenRouter, Cloudflare, and Netlify start serving Jev LangChain ships a Jev classifier, then parks it on the agent loop

docs.typesafe.ai

What Jev returns: System One in plain language

docs.typesafe.ai

OpenAI Decisions API is generally available; this desk's October 23 POSTs returned HTTP 200

docs.typesafe.ai

On eight Stripe passages, a 0-9 Score put Roll secret at 8.86 and the other six at 2.42 or below CLERC top-1 moves from 5% to 18% when jev-1.12 reranks the shortlist

docs.typesafe.ai

On eight Stripe passages, a 0-9 Score put Roll secret at 8.86 and the other six at 2.42 or below

docs.typesafe.ai

CLERC top-1 moves from 5% to 18% when jev-1.12 reranks the shortlist

evals.typesafe.ai

What Jev returns: System One in plain language

status.typesafe.ai

TypeSafe posted an API and console outage, then said it was back

status.typesafe.ai

TypeSafe posted an API and console outage, then said it was back

typesafe.ai

Every ran 777 judgments in under a second and still wanted a better accuracy check TypeSafe leaves stealth and ships Jev, a model that refuses to chat What Jev returns: System One in plain language

typesafe.ai

On October 26 TypeSafe's jev-latest named jev-1.14.0, and both aliases point at 1.14.0 TypeLLM said Jev 1.14 cannot return every vehicle box, including Tesla's flying Roadster; this desk's every-vehicle Box was one rectangle, n = 3, and type array returned HTTP 400 A green Continue button scored noul 0.99 on jev-1.14.0, n = 3, and an empty field scored 0.02 TypeSafe-Danger-Zone-Long-Context raises both Jev context caps to 262k tokens

typesafe.ai

Almeida's bet on Jevons paradox over chatbot price cuts

typesafe.ai

TypeSafe starts pulling people off the Jev waitlist TypeSafe leaves stealth and ships Jev, a model that refuses to chat

x.com

laya-mlx runs the Laya checkpoints on Apple Silicon

x.com

Qwen-Audio-Jev prints 78.70% on MMAU-mini and about 20x the base model's speed

x.com

Jevsus ran a true/false diode on jev-1.13.0, and 45% of statements move with the wording

x.com

On WANDS, more Jev questions move the mean to 0.610 and leave the median at 0.561

x.com

Abe's 1,774 Snowflake posts cost $0.055 on Jev Noul against $1.87 on AI_FILTER

x.com

Abe's 1,774 Snowflake posts cost $0.055 on Jev Noul against $1.87 on AI_FILTER

x.com

On Ámbar Reflex the 4B logit player went further than the more accurate 9B

x.com

On Ámbar Reflex the 4B logit player went further than the more accurate 9B

x.com

spark-s1-4b-v6 swaps in Qwen3.5 and prints a public-tier 83.1

x.com

On WANDS, more Jev questions move the mean to 0.610 and leave the median at 0.561

x.com

Tetris as a legal-move test: Jev picks from the landings you give it

x.com

AgentOS posts a 121-turn routing table for an opt-in Jev strategy

x.com

AgentOS posts a 121-turn routing table for an opt-in Jev strategy

x.com

AgentOS posts a 121-turn routing table for an opt-in Jev strategy

x.com

Vercel's form router keeps Jev's destination when confidence is at least 0.95

x.com

RLCDAlignBench puts a generic Noul at median AUROC 0.886, and median ECE at 0.168

x.com

On 40 client requests, Jev and Opus both scored 37 of 40, at $0.0015 against $0.005 to $0.015

x.com

On 40 client requests, Jev and Opus both scored 37 of 40, at $0.0015 against $0.005 to $0.015

x.com

On 40 client requests, Jev and Opus both scored 37 of 40, at $0.0015 against $0.005 to $0.015

x.com

On 40 client requests, Jev and Opus both scored 37 of 40, at $0.0015 against $0.005 to $0.015

x.com

On 40 client requests, Jev and Opus both scored 37 of 40, at $0.0015 against $0.005 to $0.015

x.com

On 492 tool calls, OpenRouter Armor's default Jev 1.13 scored 90.7%, 446 of 492

x.com

OpenRouter, Cloudflare, and Netlify start serving Jev

x.com

postgres-search can sink a bad hit, and the README has not measured that step

x.com

SubQ Snap 1.1 prints 65 on Artificial Analysis Decision Index, $0.04 per million input

x.com

SubQ Snap 1.1 prints 65 on Artificial Analysis Decision Index, $0.04 per million input

x.com

SubQ Snap 1.1 prints 65 on Artificial Analysis Decision Index, $0.04 per million input

x.com

SubQ Snap 1.1 prints 65 on Artificial Analysis Decision Index, $0.04 per million input

x.com

SubQ Snap 1.1 prints 65 on Artificial Analysis Decision Index, $0.04 per million input

x.com

SubQ Snap 1.1 prints 65 on Artificial Analysis Decision Index, $0.04 per million input

x.com

Almeida's bet on Jevons paradox over chatbot price cuts

x.com

OR-decision-4B prints JevBench 193 of 231, and Choice 79.7 against Jev 79.6 on alphaXiv's chart

x.com

OR-decision-4B prints JevBench 193 of 231, and Choice 79.7 against Jev 79.6 on alphaXiv's chart Base Labs posts mammoth-1 at Decision Index 0.4 Full 71.24; live 0.4.1 prints 68.4

x.com

OR-decision-4B prints JevBench 193 of 231, and Choice 79.7 against Jev 79.6 on alphaXiv's chart

x.com

OR-decision-4B prints JevBench 193 of 231, and Choice 79.7 against Jev 79.6 on alphaXiv's chart

x.com

Aman Kumar ran about 16,000 Jev calls and called it a filter, not a classifier

x.com

RLCDAlignBench puts a generic Noul at median AUROC 0.886, and median ECE at 0.168

x.com

JEV Search moves NFCorpus nDCG@10 from 0.322 to 0.377

x.com

Perplexity posted pplx-decider-v1.2 on October 16, with Decision Index 0.3 overall 68.09 on joint requests against v1.1 at 62.30

x.com

Perplexity posted pplx-decider-v1.2 on October 16, with Decision Index 0.3 overall 68.09 on joint requests against v1.1 at 62.30

x.com

Arena scores typesafe/jev-router at +3.6% and $0.155 a task

x.com

Arena scores typesafe/jev-router at +3.6% and $0.155 a task

x.com

Arena scores typesafe/jev-router at +3.6% and $0.155 a task

x.com

Arena scores typesafe/jev-router at +3.6% and $0.155 a task

x.com

Arena scores typesafe/jev-router at +3.6% and $0.155 a task

x.com

A Pi extension asks Jev which skills to load, and the 20-skill demo drops about 1,000 tokens

x.com

Opened October 16, Artificial Analysis Decision Index prints d1-ultra at 65 and Jev 1.14 at 57

x.com

Opened October 16, Artificial Analysis Decision Index prints d1-ultra at 65 and Jev 1.14 at 57 Celeris-1 Decision prints 80.0% on 3,210 jev-bench items at a 67 ms median, and Artificial Analysis lists it at 52

x.com

SubQ Snap 1.1 prints 65 on Artificial Analysis Decision Index, $0.04 per million input

x.com

Opened October 16, Artificial Analysis Decision Index prints d1-ultra at 65 and Jev 1.14 at 57

x.com

Opened October 16, Artificial Analysis Decision Index prints d1-ultra at 65 and Jev 1.14 at 57

x.com

Spring publishes a Java client that asks Jev before the chat model runs

x.com

Tetris as a legal-move test: Jev picks from the landings you give it

x.com

AstroHan ran 30 coding tasks with a Jev context filter: 25 passed, 22 without it

x.com

AstroHan ran 30 coding tasks with a Jev context filter: 25 passed, 22 without it

x.com

AstroHan ran 30 coding tasks with a Jev context filter: 25 passed, 22 without it

x.com

AstroHan ran 30 coding tasks with a Jev context filter: 25 passed, 22 without it

x.com

Opened October 16, Artificial Analysis Decision Index prints d1-ultra at 65 and Jev 1.14 at 57

x.com

AutoTrust scores GEV-26B-Decide at 62.48 on its own Decision Index run

x.com

JEV-27B-VL finishes 15 of 20 pick-and-place scenes

x.com

AutoTrust scores GEV-26B-Decide at 62.48 on its own Decision Index run

x.com

JEV-27B-VL finishes 15 of 20 pick-and-place scenes

x.com

CLERC top-1 moves from 5% to 18% when jev-1.12 reranks the shortlist

x.com

TypeLLM said Jev 1.14 cannot return every vehicle box, including Tesla's flying Roadster; this desk's every-vehicle Box was one rectangle, n = 3, and type array returned HTTP 400

x.com

On October 26 TypeSafe's jev-latest named jev-1.14.0, and both aliases point at 1.14.0 TypeLLM said Jev 1.14 cannot return every vehicle box, including Tesla's flying Roadster; this desk's every-vehicle Box was one rectangle, n = 3, and type array returned HTTP 400

x.com

On October 26 TypeSafe's jev-latest named jev-1.14.0, and both aliases point at 1.14.0 TypeLLM said Jev 1.14 cannot return every vehicle box, including Tesla's flying Roadster; this desk's every-vehicle Box was one rectangle, n = 3, and type array returned HTTP 400 A green Continue button scored noul 0.99 on jev-1.14.0, n = 3, and an empty field scored 0.02

x.com

A week of Jev repos: routers, MCP, graphs, games

x.com

RLCDAlignBench puts a generic Noul at median AUROC 0.886, and median ECE at 0.168

x.com

Base Labs posts mammoth-1 at Decision Index 0.4 Full 71.24; live 0.4.1 prints 68.4

x.com

Base Labs posts mammoth-1 at Decision Index 0.4 Full 71.24; live 0.4.1 prints 68.4

x.com

Base Labs posts mammoth-1 at Decision Index 0.4 Full 71.24; live 0.4.1 prints 68.4

x.com

Base Labs posts mammoth-1 at Decision Index 0.4 Full 71.24; live 0.4.1 prints 68.4

x.com

Base Labs posts mammoth-1 at Decision Index 0.4 Full 71.24; live 0.4.1 prints 68.4

x.com

JevBench v1.5 ranks Cygnet 73.7 and Jev 1.13.0 72.1

x.com

JevBench v1.5 ranks Cygnet 73.7 and Jev 1.13.0 72.1

x.com

JevBench v1.5 ranks Cygnet 73.7 and Jev 1.13.0 72.1

x.com

JevBench v1.5 ranks Cygnet 73.7 and Jev 1.13.0 72.1

x.com

Bespoke Nimble 9B matches 90% of its own labels and 74.8% of a public suite

x.com

Bespoke Nimble 9B matches 90% of its own labels and 74.8% of a public suite

x.com

JevBench ranks Sage 1.3.0 first of 27 hosted systems at composite 74.0

x.com

TypeLLM said Jev 1.14 cannot return every vehicle box, including Tesla's flying Roadster; this desk's every-vehicle Box was one rectangle, n = 3, and type array returned HTTP 400 A green Continue button scored noul 0.99 on jev-1.14.0, n = 3, and an empty field scored 0.02

x.com

Bud Decision Studio serves eleven local decision models on port 8420

x.com

Eikos-27B prints 82.9 on its JevBench hard row, and 2.4% error once confidence is 0.90

x.com

Eikos-27B prints 82.9 on its JevBench hard row, and 2.4% error once confidence is 0.90

x.com

Eikos-27B prints 82.9 on its JevBench hard row, and 2.4% error once confidence is 0.90

x.com

Eikos-27B prints 82.9 on its JevBench hard row, and 2.4% error once confidence is 0.90

x.com

Eikos-27B prints 82.9 on its JevBench hard row, and 2.4% error once confidence is 0.90

x.com

Hono's author routes HTTP by meaning, then warns you not to

x.com

Celeris-1 Decision prints 80.0% on 3,210 jev-bench items at a 67 ms median, and Artificial Analysis lists it at 52

x.com

Celeris-1 Decision prints 80.0% on 3,210 jev-bench items at a 67 ms median, and Artificial Analysis lists it at 52

x.com

Celeris-1 Decision prints 80.0% on 3,210 jev-bench items at a 67 ms median, and Artificial Analysis lists it at 52

x.com

Celeris-1 Decision prints 80.0% on 3,210 jev-bench items at a 67 ms median, and Artificial Analysis lists it at 52

x.com

Base Labs posts mammoth-1 at Decision Index 0.4 Full 71.24; live 0.4.1 prints 68.4

x.com

Charly Poly put Jev next to open encoders, not GPT-5.6, and it lost on calibration

x.com

Jevonian asks Jev which model should take the next coding turn

x.com

llamacpp-jev answers /v1/systemone from an unmodified llama-server, including images

x.com

DiffusionGemma as Jev reads one denoising step, and a hand chart sums to 198 of 201

x.com

Spring publishes a Java client that asks Jev before the chat model runs

x.com

Arena scores typesafe/jev-router at +3.6% and $0.155 a task Jared Palmer's Kev family copies Jev's API on Qwen, and trails it on new sources OpenRouter, Cloudflare, and Netlify start serving Jev

x.com

Cloudflare posts Clef and Clef-flash, Apache 2.0 decision models on Workers AI

x.com

Clef-omni scores a 21-second video in about 1.5 seconds, at $0.15 per million input tokens Cloudflare posts Clef and Clef-flash, Apache 2.0 decision models on Workers AI

x.com

A v3 judge paper puts a frozen Jev cascade 0.9 points above GPT-6 on 1,610 pairs

x.com

On October 26 TypeSafe's jev-latest named jev-1.14.0, and both aliases point at 1.14.0

x.com

Sthānika's Sieve-31B card prints Decision Index 0.3.2 Full score 63.8

x.com

swift-jev types Choice, Noul, and Score, and routes at 0.60 and 0.85

x.com

Jeva-4B prints 0 flips on 600 Choice permutations, and BANKING77 at K=16 is 91.62 against Jev 82.50

x.com

On eight Stripe passages, a 0-9 Score put Roll secret at 8.86 and the other six at 2.42 or below

x.com

A Cloud Run recipe serves DiffusionGemma-Jev at 35 to 60 ms

x.com

On eight Stripe passages, a 0-9 Score put Roll secret at 8.86 and the other six at 2.42 or below

x.com

Databricks posts ai_decide, a Beta function that returns a noul, a choice, or a score

x.com

Databricks posts ai_decide, a Beta function that returns a noul, a choice, or a score

x.com

Maincode's Matilda card prints 59.26 on a self-run of Decision Index 0.2.1

x.com

TypeLLM said Jev 1.14 cannot return every vehicle box, including Tesla's flying Roadster; this desk's every-vehicle Box was one rectangle, n = 3, and type array returned HTTP 400

x.com

Perplexity posted pplx-decider-v1.2 on October 16, with Decision Index 0.3 overall 68.09 on joint requests against v1.1 at 62.30

x.com

llamacpp-jev answers /v1/systemone from an unmodified llama-server, including images

x.com

TypeSafe reopens Jev signups and drops the free credit for new accounts

x.com

TypeSafe reopens Jev signups and drops the free credit for new accounts

x.com

jevgrep solves the same 8 of 10 SWE-bench tasks, and the bill including Jev is 25.8% lower

x.com

jevgrep solves the same 8 of 10 SWE-bench tasks, and the bill including Jev is 25.8% lower

x.com

On eight Stripe passages, a 0-9 Score put Roll secret at 8.86 and the other six at 2.42 or below

x.com

One day of grant posts, 100 of them, hid 75 of 78 rejects after the questions changed

x.com

One day of grant posts, 100 of them, hid 75 of 78 rejects after the questions changed

x.com

Exa posts a demo that puts web results into Jev's state

x.com

On eight Stripe passages, a 0-9 Score put Roll secret at 8.86 and the other six at 2.42 or below

x.com

TypeSafe leaves stealth and ships Jev, a model that refuses to chat What Jev returns: System One in plain language

x.com

Bespoke Nimble 9B matches 90% of its own labels and 74.8% of a public suite

x.com

On October 26 TypeSafe's jev-latest named jev-1.14.0, and both aliases point at 1.14.0

x.com

On October 26 TypeSafe's jev-latest named jev-1.14.0, and both aliases point at 1.14.0

x.com

On October 26 TypeSafe's jev-latest named jev-1.14.0, and both aliases point at 1.14.0

x.com

A green Continue button scored noul 0.99 on jev-1.14.0, n = 3, and an empty field scored 0.02

x.com

On October 26 TypeSafe's jev-latest named jev-1.14.0, and both aliases point at 1.14.0

x.com

On 492 tool calls, OpenRouter Armor's default Jev 1.13 scored 90.7%, 446 of 492

x.com

OpenRouter, Cloudflare, and Netlify start serving Jev

x.com

TypeSafe reopens Jev signups and drops the free credit for new accounts

x.com

Exa posts a demo that puts web results into Jev's state

x.com

OpenAI's Decisions API preview says results come back in 150 milliseconds

x.com

TypeSafe-Danger-Zone-Long-Context raises both Jev context caps to 262k tokens

x.com

TypeSafe reopens Jev signups and drops the free credit for new accounts

x.com

On October 26 TypeSafe's jev-latest named jev-1.14.0, and both aliases point at 1.14.0

x.com

TypeLLM said Jev 1.14 cannot return every vehicle box, including Tesla's flying Roadster; this desk's every-vehicle Box was one rectangle, n = 3, and type array returned HTTP 400

x.com

THX-01 averages 98.4 on its 2,843 support tickets

x.com

DSPy 3.4.0 calls Jev from Predict, with a gate at 0.2 and 0.6

x.com

DSPy 3.4.0 calls Jev from Predict, with a gate at 0.2 and 0.6

x.com

One post scores Jev at 100 and Laya's English checkpoint at 92, on 100 items

x.com

THX-01 averages 98.4 on its 2,843 support tickets

x.com

You.com's risk sweeps cost $0.14 each, and the report means tie

x.com

A green Continue button scored noul 0.99 on jev-1.14.0, n = 3, and an empty field scored 0.02

x.com

A 100,000-line needle takes 18.0 minutes with Jev and 24.7 without

x.com

TypeLLM said Jev 1.14 cannot return every vehicle box, including Tesla's flying Roadster; this desk's every-vehicle Box was one rectangle, n = 3, and type array returned HTTP 400

x.com

Jev scored 384 morning headlines in 25 seconds; Opus 5 finished four

x.com

PageIndex walks long filings with Jev Choice calls and checks two answers

x.com

GLiDE v1.2 prints Decision Index 0.4.1 Full 71.5, first of 229, at $0.04

x.com

GLiDE v1.2 prints Decision Index 0.4.1 Full 71.5, first of 229, at $0.04

x.com

SimpleJev turns an open model into a classifier by reading the next-token logits

x.com

Opened October 17, JevBench API board v1.7.46 ranks kettei-fat first of 39 at composite 77.8

x.com

Opened October 17, JevBench API board v1.7.46 ranks kettei-fat first of 39 at composite 77.8

x.com

jevos-1b prints 0.815 on 2,000 yes/no questions, against Jev at 0.927

x.com

jevos-1b prints 0.815 on 2,000 yes/no questions, against Jev at 0.927

x.com

jevos-1b prints 0.815 on 2,000 yes/no questions, against Jev at 0.927

x.com

TechCrunch covers Jev, and quotes a Vercel engineer who swapped Luna for it

x.com

JevBench v1.5 ranks Cygnet 73.7 and Jev 1.13.0 72.1

x.com

Opened October 17, JevBench API board v1.7.46 ranks kettei-fat first of 39 at composite 77.8

x.com

Decision Index 0.2 prints Jev at 51.67, and a later post names another first place JevBench v1.5 ranks Cygnet 73.7 and Jev 1.13.0 72.1

x.com

Plumb-4B reaches 89 of 111 public hard items, and the official board scored it 65.84 JevBench v1.5 ranks Cygnet 73.7 and Jev 1.13.0 72.1

x.com

GLiDE v1.2 prints Decision Index 0.4.1 Full 71.5, first of 229, at $0.04

x.com

JevBench v1.5 ranks Cygnet 73.7 and Jev 1.13.0 72.1 SemIf, a 4B open rebuild that reads option logits, sits 0.7 behind Jev on JevBench

x.com

djev reads DiffusionGemma as a decision API, and sits third on JevBench SimpleJev turns an open model into a classifier by reading the next-token logits

x.com

Jared Palmer's Kev family copies Jev's API on Qwen, and trails it on new sources

x.com

Bespoke Nimble 9B matches 90% of its own labels and 74.8% of a public suite

x.com

JevBench ranks Sage 1.3.0 first of 27 hosted systems at composite 74.0 Microsoft-Decision-1 lists $0.042 per million input tokens and calls itself 4.5 times quicker than Quyet-1.0-Large JevBench v1.5 ranks Cygnet 73.7 and Jev 1.13.0 72.1

x.com

JevBench v1.5 ranks Cygnet 73.7 and Jev 1.13.0 72.1

x.com

JevBench v1.5 ranks Cygnet 73.7 and Jev 1.13.0 72.1

x.com

JevBench v1.5 ranks Cygnet 73.7 and Jev 1.13.0 72.1

x.com

Quyet-1.0-Large's homepage prints JevBench v1.6.0 at 71.39

x.com

JevBench v1.5 ranks Cygnet 73.7 and Jev 1.13.0 72.1

x.com

Browser Use wires Jev into the click loop: 7 seconds, $0.0039

x.com

A week of Jev repos: routers, MCP, graphs, games

x.com

fsdecide scores 91.3% on ForkScape review, and Jev scores 56.3%

x.com

OpenRouter, Cloudflare, and Netlify start serving Jev

x.com

Blink v0.3 prints Decision Index 0.2.1 at 57.48

x.com

Blink v0.3 prints Decision Index 0.2.1 at 57.48

x.com

NIRNAY 450M prints 0.8792 on 3,080 Banking77 rows and says the Jev cell was not measured here

x.com

NIRNAY 450M prints 0.8792 on 3,080 Banking77 rows and says the Jev cell was not measured here

x.com

NIRNAY 450M prints 0.8792 on 3,080 Banking77 rows and says the Jev cell was not measured here

x.com

NIRNAY 450M prints 0.8792 on 3,080 Banking77 rows and says the Jev cell was not measured here

x.com

NIRNAY 450M prints 0.8792 on 3,080 Banking77 rows and says the Jev cell was not measured here

x.com

Rizzo Flow is 49 ms on an RTX 5060 Ti, and six missing-evidence rows stay confident

x.com

GLiDE v1.2 prints Decision Index 0.4.1 Full 71.5, first of 229, at $0.04

x.com

GLiDE v1.2 prints Decision Index 0.4.1 Full 71.5, first of 229, at $0.04

x.com

GLiDE v1.2 prints Decision Index 0.4.1 Full 71.5, first of 229, at $0.04

x.com

GitHub Next ships LocalJev, a prompted stand-in that speaks /v1/systemone

x.com

On October 26 TypeSafe's jev-latest named jev-1.14.0, and both aliases point at 1.14.0

x.com

A Cloud Run recipe serves DiffusionGemma-Jev at 35 to 60 ms

x.com

A Cloud Run recipe serves DiffusionGemma-Jev at 35 to 60 ms

x.com

DiffusionGemma as Jev reads one denoising step, and a hand chart sums to 198 of 201

x.com

DiffusionGemma as Jev reads one denoising step, and a hand chart sums to 198 of 201

x.com

DiffusionGemma as Jev reads one denoising step, and a hand chart sums to 198 of 201

x.com

Browser Use wires Jev into the click loop: 7 seconds, $0.0039

x.com

OpenAI Decisions API is generally available; this desk's October 23 POSTs returned HTTP 200

x.com

Opened October 17, JevBench API board v1.7.46 ranks kettei-fat first of 39 at composite 77.8

x.com

Span-02's card prints Decision Index 0.3 text index 63.62, and the blog prints thinking off at 61.2 against Jev 58.0 Perplexity posted pplx-decider-v1.2 on October 16, with Decision Index 0.3 overall 68.09 on joint requests against v1.1 at 62.30

x.com

Span-02's card prints Decision Index 0.3 text index 63.62, and the blog prints thinking off at 61.2 against Jev 58.0

x.com

Liquid's chart puts d1 at 58.9 on an internal Decision Index 0.2.1

x.com

HydroJEV tests jev-1.13.0 as a one-second screen on simulated water-network alarms

x.com

Bespoke Nimble 9B matches 90% of its own labels and 74.8% of a public suite

x.com

Span-02's card prints Decision Index 0.3 text index 63.62, and the blog prints thinking off at 61.2 against Jev 58.0

x.com

Decision Index 0.2 prints Jev at 51.67, and a later post names another first place

x.com

Almeida's bet on Jevons paradox over chatbot price cuts

x.com

OpenAI Decisions API is generally available; this desk's October 23 POSTs returned HTTP 200

x.com

OpenAI Decisions API is generally available; this desk's October 23 POSTs returned HTTP 200

x.com

dejevu books the same Flights search faster than the published Jev demo, on another machine

x.com

Inception calls Mercury Decide first on OpenRouter's JevBench v1.4, and prints no score

x.com

JevAdvBench's abstract says one appended opinion flips 12.1% of jev-1.13.0 decisions

x.com

Exa posts a demo that puts web results into Jev's state

x.com

On October 26 TypeSafe's jev-latest named jev-1.14.0, and both aliases point at 1.14.0

x.com

SREGym put Jev on a Codex agent: 24 passes out of 50, up from 20

x.com

CLM-8B caches the action list, and a fine-tuned head scores 31 of 38 DeepSWE holdouts

x.com

CLM-8B caches the action list, and a fine-tuned head scores 31 of 38 DeepSWE holdouts

x.com

CLM-8B caches the action list, and a fine-tuned head scores 31 of 38 DeepSWE holdouts

x.com

On 492 tool calls, OpenRouter Armor's default Jev 1.13 scored 90.7%, 446 of 492

x.com

StreamDecisionBench puts Kev-27B at 62.15 and Jev at 59.63 Jared Palmer's Kev family copies Jev's API on Qwen, and trails it on new sources

x.com

StreamDecisionBench puts Kev-27B at 62.15 and Jev at 59.63

x.com

jevctl puts Jev on the command line, with a compaction hook that keeps lines verbatim

x.com

Jevis is a Flutter integration-test agent that only taps what you registered

x.com

Five runs leave a 3.3-point lead on 18 claim checks

x.com

Five runs leave a 3.3-point lead on 18 claim checks

x.com

Jared Palmer's Kev family copies Jev's API on Qwen, and trails it on new sources

x.com

Jared Palmer's Kev family copies Jev's API on Qwen, and trails it on new sources

x.com

Jared Palmer's Kev family copies Jev's API on Qwen, and trails it on new sources

x.com

Jared Palmer's Kev family copies Jev's API on Qwen, and trails it on new sources

x.com

Jared Palmer's Kev family copies Jev's API on Qwen, and trails it on new sources

x.com

Jared Palmer's Kev family copies Jev's API on Qwen, and trails it on new sources

x.com

A buy-or-sell every Monad block, with a mock default model

x.com

Compaction without summarization: score the tool calls and drop them

x.com

Jason Lemkin tried Jev on SaaStr Connect and stopped on a 32% false-admission rate

x.com

On DeepEval's Paris weather example, jev-1.13.0 scored 0.254 when the answer invented a breeze

x.com

On DeepEval's Paris weather example, jev-1.13.0 scored 0.254 when the answer invented a breeze

x.com

On DeepEval's Paris weather example, jev-1.13.0 scored 0.254 when the answer invented a breeze

x.com

DocJev classifies and splits real PDFs with Jev, 5.7x to 6.5x faster than Luna on the author's set

x.com

DocJev classifies and splits real PDFs with Jev, 5.7x to 6.5x faster than Luna on the author's set

x.com

JevPilot is a playable driving sim that asks Jev which path to take

x.com

jsort ranks a file by asking Jev which of two texts is more of something

x.com

jsort ranks a file by asking Jev which of two texts is more of something

x.com

Stagehand, Chrome, and a11y trees: computer use as a Jev question

x.com

Wrangle's chart prints 137.3 perfect matches per 200 with Jev

x.com

On eight Stripe passages, a 0-9 Score put Roll secret at 8.86 and the other six at 2.42 or below

x.com

pg-jev puts prompt_jev() in PostgreSQL and returns jsonb

x.com

DiffusionGemma as Jev reads one denoising step, and a hand chart sums to 198 of 201

x.com

Jev macOS Loop clicks native Mac windows; Finder sorted nine files in 7.39 seconds

x.com

Jeva-4B prints 0 flips on 600 Choice permutations, and BANKING77 at K=16 is 91.62 against Jev 82.50

x.com

JevPilot is a playable driving sim that asks Jev which path to take

x.com

A green Continue button scored noul 0.99 on jev-1.14.0, n = 3, and an empty field scored 0.02

x.com

The Jeeves README prints 0.865 on the public JevBench hard tier

x.com

OpenAI Decisions API is generally available; this desk's October 23 POSTs returned HTTP 200

x.com

Cloudflare posts Clairic, a System One extraction model at $0.10 per million input

x.com

TypeSafe starts pulling people off the Jev waitlist

x.com

TypeSafe reopens Jev signups and drops the free credit for new accounts

x.com

jsort ranks a file by asking Jev which of two texts is more of something

x.com

jsort ranks a file by asking Jev which of two texts is more of something

x.com

An n8n community node routes Jev answers onto category branches, including Needs Review

x.com

llamacpp-jev answers /v1/systemone from an unmodified llama-server, including images

x.com

llamacpp-jev answers /v1/systemone from an unmodified llama-server, including images

x.com

llamacpp-jev answers /v1/systemone from an unmodified llama-server, including images

x.com

llamacpp-jev answers /v1/systemone from an unmodified llama-server, including images

x.com

Aman Kumar ran about 16,000 Jev calls and called it a filter, not a classifier

x.com

Stagehand, Chrome, and a11y trees: computer use as a Jev question

x.com

jevsearch re-ranks keyword hits with Jev and scored 83% Hit@1 on the author's TypeSafe docs set

x.com

jevsearch re-ranks keyword hits with Jev and scored 83% Hit@1 on the author's TypeSafe docs set

x.com

LangChain ran Jev as a judge: 500 matching pass/fail labels, and a $0.34 bill against Claude's $28

x.com

meraGPT posts Decider 1 ahead of Jev on a teacher-labelled set

x.com

meraGPT posts Decider 1 ahead of Jev on a teacher-labelled set

x.com

meraGPT posts Decider 1 ahead of Jev on a teacher-labelled set

x.com

meraGPT posts Decider 1 ahead of Jev on a teacher-labelled set

x.com

Step-Jev scores 19.6% when the reward grades every search step

x.com

Step-Jev scores 19.6% when the reward grades every search step

x.com

Step-Jev scores 19.6% when the reward grades every search step

x.com

Step-Jev scores 19.6% when the reward grades every search step

x.com

Step-Jev scores 19.6% when the reward grades every search step

x.com

Lightpanda's hostel script counts browser memory, and the Jev clock stays in the post

x.com

Lightpanda's hostel script counts browser memory, and the Jev clock stays in the post

x.com

meraGPT posts Decider 1 ahead of Jev on a teacher-labelled set

x.com

Opened October 16, Artificial Analysis Decision Index prints d1-ultra at 65 and Jev 1.14 at 57

x.com

Opened October 16, Artificial Analysis Decision Index prints d1-ultra at 65 and Jev 1.14 at 57

x.com

Opened October 16, Artificial Analysis Decision Index prints d1-ultra at 65 and Jev 1.14 at 57

x.com

Open d1's d1-3B card prints 48.57 on a self-run of Decision Index 0.2.1

x.com

Liquid's chart puts d1 at 58.9 on an internal Decision Index 0.2.1

x.com

Open d1's d1-3B card prints 48.57 on a self-run of Decision Index 0.2.1

x.com

Open d1's d1-3B card prints 48.57 on a self-run of Decision Index 0.2.1

x.com

Open d1's d1-3B card prints 48.57 on a self-run of Decision Index 0.2.1

x.com

Opened October 16, Artificial Analysis Decision Index prints d1-ultra at 65 and Jev 1.14 at 57

x.com

Opened October 16, Artificial Analysis Decision Index prints d1-ultra at 65 and Jev 1.14 at 57

x.com

Open d1's d1-3B card prints 48.57 on a self-run of Decision Index 0.2.1

x.com

Open d1's d1-3B card prints 48.57 on a self-run of Decision Index 0.2.1 Liquid's chart puts d1 at 58.9 on an internal Decision Index 0.2.1

x.com

Opened October 16, Artificial Analysis Decision Index prints d1-ultra at 65 and Jev 1.14 at 57

x.com

Liquid's chart puts d1 at 58.9 on an internal Decision Index 0.2.1 Decision Index 0.2 prints Jev at 51.67, and a later post names another first place

x.com

Opened October 16, Artificial Analysis Decision Index prints d1-ultra at 65 and Jev 1.14 at 57

x.com

LiteLLM adds day-0 Jev 1.14 images, box questions, and long context on both decision routes

x.com

LlamaIndex's Logan Markewich ships jeff, a local System One API on GLiFormer

x.com

JevBench v1.5 ranks Cygnet 73.7 and Jev 1.13.0 72.1

x.com

Laya, an Apache 2.0 System One stack, publishes where it beats Jev's numbers and where it does not

x.com

astra-jev-harness cuts Astra input 27% on one two-file fixture

x.com

astra-jev-harness cuts Astra input 27% on one two-file fixture

x.com

Jevflake puts Jev in Snowflake: noul, choice, and score as SQL functions

x.com

Bespoke Nimble 9B matches 90% of its own labels and 74.8% of a public suite

x.com

Bespoke Nimble 9B matches 90% of its own labels and 74.8% of a public suite

x.com

Bespoke Nimble 9B matches 90% of its own labels and 74.8% of a public suite

x.com

Maincode's Matilda card prints 59.26 on a self-run of Decision Index 0.2.1 Decision Index 0.2 prints Jev at 51.67, and a later post names another first place

x.com

Almeida's bet on Jevons paradox over chatbot price cuts

x.com

Mizan 4B averages 78.7 on 3,000 Persian records

x.com

Mizan 4B averages 78.7 on 3,000 Persian records

x.com

Mizan 4B averages 78.7 on 3,000 Persian records

x.com

Opened October 26, Decision Index 0.4.1 Full score for d3-ultra is 70.18, second of 229

x.com

DiffusionGemma as Jev reads one denoising step, and a hand chart sums to 198 of 201

x.com

hunch runs EmbeddingGemma2 as a CPU System One endpoint at 20 ms on text and 100 ms on images

x.com

djev reads DiffusionGemma as a decision API, and sits third on JevBench

x.com

hunch runs EmbeddingGemma2 as a CPU System One endpoint at 20 ms on text and 100 ms on images

x.com

hunch runs EmbeddingGemma2 as a CPU System One endpoint at 20 ms on text and 100 ms on images

x.com

hunch runs EmbeddingGemma2 as a CPU System One endpoint at 20 ms on text and 100 ms on images

x.com

hunch runs EmbeddingGemma2 as a CPU System One endpoint at 20 ms on text and 100 ms on images

x.com

DiffusionGemma as Jev reads one denoising step, and a hand chart sums to 198 of 201

x.com

DiffusionGemma as Jev reads one denoising step, and a hand chart sums to 198 of 201

x.com

hunch runs EmbeddingGemma2 as a CPU System One endpoint at 20 ms on text and 100 ms on images

x.com

hunch runs EmbeddingGemma2 as a CPU System One endpoint at 20 ms on text and 100 ms on images

x.com

TypeSafe reopens Jev signups and drops the free credit for new accounts

x.com

jev-rubiks asks Jev when to speak, after a test where picking moves did nothing

x.com

On DeepEval's Paris weather example, jev-1.13.0 scored 0.254 when the answer invented a breeze

x.com

A hybrid Jev overlay sped up Postgres joins 12% on JOB, and made the planner slower

x.com

hotdog-27B leads a 1,631-decision holdout and trails on 74 public items

x.com

Cloudflare posts Clairic, a System One extraction model at $0.10 per million input

x.com

Cloudflare posts Clairic, a System One extraction model at $0.10 per million input

x.com

Cloudflare posts Clairic, a System One extraction model at $0.10 per million input

x.com

Cloudflare posts Clef and Clef-flash, Apache 2.0 decision models on Workers AI

x.com

Cloudflare posts Clairic, a System One extraction model at $0.10 per million input

x.com

A chart puts Jev's reading of 25 probability words next to 99 people

x.com

A chart puts Jev's reading of 25 probability words next to 99 people

x.com

A gradient-boosting ensemble beats Jev 1.13 on five CYP assays

x.com

jgrep's test bench prints mean recall 0.926 across 60 commits, at $0.1112

x.com

mizchi's fit notes record 0 illegal moves in 489 MOBA calls, and 6 to 27 wrong pages where a one-line rule was 16/16

x.com

semgate asks Jev about each HTTP request, then a live range to attack it

x.com

semgate asks Jev about each HTTP request, then a live range to attack it

x.com

imajev-4b: Hugging Face weights and JevBench 70.4

x.com

imajev-4b: Hugging Face weights and JevBench 70.4

x.com

imajev-4b: Hugging Face weights and JevBench 70.4

x.com

TypeSafe reopens Jev signups and drops the free credit for new accounts

x.com

MotherDuck's prompt_jev() labels 100,000 AG News rows in 40 seconds for $0.50

x.com

Decision Index 0.2 prints Jev at 51.67, and a later post names another first place

x.com

Bryo ran 1,565 business emails: Jev at 96.4%, Gemini a point or two higher

x.com

Bryo ran 1,565 business emails: Jev at 96.4%, Gemini a point or two higher

x.com

Bryo ran 1,565 business emails: Jev at 96.4%, Gemini a point or two higher

x.com

Bryo ran 1,565 business emails: Jev at 96.4%, Gemini a point or two higher

x.com

ASO Workbench ranks App Store terms with a model trained on Jev's answers

x.com

Nace posts Drex Max at Decision Index 0.3.2 public 67.59, $0.06 per million input

x.com

Nace posts Drex Max at Decision Index 0.3.2 public 67.59, $0.06 per million input

x.com

Nace posts Drex 1.5 at $0.04, and its chart shows 199 of 231 against Jev's 201

x.com

Nace posts Drex Max at Decision Index 0.3.2 public 67.59, $0.06 per million input

x.com

Nace posts Drex 1.1, and the card titles the weights Drex DLM Nace posts Drex 1.5 at $0.04, and its chart shows 199 of 231 against Jev's 201

x.com

Nace posts Drex 1.5 at $0.04, and its chart shows 199 of 231 against Jev's 201 Decision Index 0.2 prints Jev at 51.67, and a later post names another first place

x.com

Nace posts Drex Max at Decision Index 0.3.2 public 67.59, $0.06 per million input

x.com

Nace posts Drex Max at Decision Index 0.3.2 public 67.59, $0.06 per million input

x.com

Nace posts Drex Max at Decision Index 0.3.2 public 67.59, $0.06 per million input

x.com

Nace posts Drex 1.5 at $0.04, and its chart shows 199 of 231 against Jev's 201

x.com

Nace posts Drex 1.1, and the card titles the weights Drex DLM

x.com

Vega, the vegaml package, scores 75.4% on 800 phishing emails against Jev at 61.9%

x.com

Vega, the vegaml package, scores 75.4% on 800 phishing emails against Jev at 61.9%

x.com

TypeLLM said Jev 1.14 cannot return every vehicle box, including Tesla's flying Roadster; this desk's every-vehicle Box was one rectangle, n = 3, and type array returned HTTP 400

x.com

TypeLLM said Jev 1.14 cannot return every vehicle box, including Tesla's flying Roadster; this desk's every-vehicle Box was one rectangle, n = 3, and type array returned HTTP 400

x.com

On eight Stripe passages, a 0-9 Score put Roll secret at 8.86 and the other six at 2.42 or below

x.com

On eight Stripe passages, a 0-9 Score put Roll secret at 8.86 and the other six at 2.42 or below

x.com

On eight Stripe passages, a 0-9 Score put Roll secret at 8.86 and the other six at 2.42 or below

x.com

On WANDS, more Jev questions move the mean to 0.610 and leave the median at 0.561

x.com

On WANDS, more Jev questions move the mean to 0.610 and leave the median at 0.561

x.com

On eight Stripe passages, a 0-9 Score put Roll secret at 8.86 and the other six at 2.42 or below

x.com

DSPy 3.4.0 calls Jev from Predict, with a gate at 0.2 and 0.6

x.com

BranchOne returns Choice probabilities from Qwen3.5-0.8B-Base, median 0.305 s on ten short text questions

x.com

BranchOne returns Choice probabilities from Qwen3.5-0.8B-Base, median 0.305 s on ten short text questions

x.com

OpenRouter, Cloudflare, and Netlify start serving Jev

x.com

OR-decision-4B prints JevBench 193 of 231, and Choice 79.7 against Jev 79.6 on alphaXiv's chart

x.com

A registered attack on jev-1.13.0 moves validated success from 1.8% to 3.5%

x.com

Datadog prints Jev Noul recall 82.90% to 3.88% under a counterfeit policy, and this desk's is_risky fell from 0.94 to 0.13

x.com

On eight Stripe passages, a 0-9 Score put Roll secret at 8.86 and the other six at 2.42 or below

x.com

On eight Stripe passages, a 0-9 Score put Roll secret at 8.86 and the other six at 2.42 or below

x.com

CLERC top-1 moves from 5% to 18% when jev-1.12 reranks the shortlist

x.com

TechCrunch covers Jev, and quotes a Vercel engineer who swapped Luna for it Bryo ran 1,565 business emails: Jev at 96.4%, Gemini a point or two higher

x.com

Nace posts Drex Max at Decision Index 0.3.2 public 67.59, $0.06 per million input

x.com

Opened October 17, JevBench API board v1.7.46 ranks kettei-fat first of 39 at composite 77.8

x.com

A frozen Jev pipeline reports the lowest ten-corpus valence-arousal error in a SemEval table

x.com

JevGate turns one function at a time into a finding at 0.80

x.com

PurpleMIST-Flash-1.0 scores 0.668 on 15,050 translated African-language decisions

x.com

PurpleMIST-Flash-1.0 scores 0.668 on 15,050 translated African-language decisions

x.com

Jared Palmer's Kev family copies Jev's API on Qwen, and trails it on new sources

x.com

Bespoke Nimble 9B matches 90% of its own labels and 74.8% of a public suite

x.com

Bespoke Nimble 9B matches 90% of its own labels and 74.8% of a public suite

x.com

Almeida's bet on Jevons paradox over chatbot price cuts

x.com

A buy-or-sell every Monad block, with a mock default model

x.com

OpenAI Decisions API is generally available; this desk's October 23 POSTs returned HTTP 200 OpenAI's Decisions API preview says results come back in 150 milliseconds

x.com

OpenAI's Decisions API preview says results come back in 150 milliseconds

x.com

OpenAI Decisions API is generally available; this desk's October 23 POSTs returned HTTP 200 OpenAI's Decisions API preview says results come back in 150 milliseconds

x.com

OpenAI Decisions API is generally available; this desk's October 23 POSTs returned HTTP 200 OpenAI's Decisions API preview says results come back in 150 milliseconds

x.com

OpenAI Decisions API is generally available; this desk's October 23 POSTs returned HTTP 200

x.com

OpenAI Decisions API is generally available; this desk's October 23 POSTs returned HTTP 200 OpenAI's Decisions API preview says results come back in 150 milliseconds

x.com

MiniCPM5-2B-Jev prints 78.79% on JevBench, first among open models at or under 2B

x.com

OpenRouter, Cloudflare, and Netlify start serving Jev

x.com

Arena scores typesafe/jev-router at +3.6% and $0.155 a task OpenRouter, Cloudflare, and Netlify start serving Jev

x.com

On 492 tool calls, OpenRouter Armor's default Jev 1.13 scored 90.7%, 446 of 492

x.com

On 492 tool calls, OpenRouter Armor's default Jev 1.13 scored 90.7%, 446 of 492

x.com

Perplexity posted pplx-decider-v1.2 on October 16, with Decision Index 0.3 overall 68.09 on joint requests against v1.1 at 62.30

x.com

Arena scores typesafe/jev-router at +3.6% and $0.155 a task OpenRouter, Cloudflare, and Netlify start serving Jev

x.com

On October 26 TypeSafe's jev-latest named jev-1.14.0, and both aliases point at 1.14.0

x.com

Jared Palmer's Kev family copies Jev's API on Qwen, and trails it on new sources

x.com

Arena scores typesafe/jev-router at +3.6% and $0.155 a task

x.com

On 492 tool calls, OpenRouter Armor's default Jev 1.13 scored 90.7%, 446 of 492

x.com

Microsoft-Decision-1 lists $0.042 per million input tokens and calls itself 4.5 times quicker than Quyet-1.0-Large

x.com

On 492 tool calls, OpenRouter Armor's default Jev 1.13 scored 90.7%, 446 of 492

x.com

Arena scores typesafe/jev-router at +3.6% and $0.155 a task

x.com

Arena scores typesafe/jev-router at +3.6% and $0.155 a task

x.com

Inception calls Mercury Decide first on OpenRouter's JevBench v1.4, and prints no score

x.com

Arena scores typesafe/jev-router at +3.6% and $0.155 a task

x.com

Arena scores typesafe/jev-router at +3.6% and $0.155 a task

x.com

On WANDS, more Jev questions move the mean to 0.610 and leave the median at 0.561

x.com

PageIndex walks long filings with Jev Choice calls and checks two answers

x.com

jevelry turns a Jev call into a folder, a verdict, and a log

x.com

On October 26 TypeSafe's jev-latest named jev-1.14.0, and both aliases point at 1.14.0

x.com

Perplexity posted pplx-decider-v1.2 on October 16, with Decision Index 0.3 overall 68.09 on joint requests against v1.1 at 62.30

x.com

Perplexity posted pplx-decider-v1.2 on October 16, with Decision Index 0.3 overall 68.09 on joint requests against v1.1 at 62.30

x.com

Perplexity posted pplx-decider-v1.2 on October 16, with Decision Index 0.3 overall 68.09 on joint requests against v1.1 at 62.30

x.com

Perplexity posted pplx-decider-v1.2 on October 16, with Decision Index 0.3 overall 68.09 on joint requests against v1.1 at 62.30

x.com

Perplexity posted pplx-decider-v1.2 on October 16, with Decision Index 0.3 overall 68.09 on joint requests against v1.1 at 62.30

x.com

Perplexity posted pplx-decider-v1.2 on October 16, with Decision Index 0.3 overall 68.09 on joint requests against v1.1 at 62.30 Decision Index 0.2 prints Jev at 51.67, and a later post names another first place

x.com

Perplexity posted pplx-decider-v1.2 on October 16, with Decision Index 0.3 overall 68.09 on joint requests against v1.1 at 62.30

x.com

A 612-row banking-feed test puts Jev under two Gemini rows on accuracy

x.com

Charly Poly put Jev next to open encoders, not GPT-5.6, and it lost on calibration

x.com

Charly Poly put Jev next to open encoders, not GPT-5.6, and it lost on calibration

x.com

Charly Poly put Jev next to open encoders, not GPT-5.6, and it lost on calibration

x.com

Charly Poly put Jev next to open encoders, not GPT-5.6, and it lost on calibration

x.com

Charly Poly put Jev next to open encoders, not GPT-5.6, and it lost on calibration

x.com

Charly Poly put Jev next to open encoders, not GPT-5.6, and it lost on calibration

x.com

Charly Poly put Jev next to open encoders, not GPT-5.6, and it lost on calibration

x.com

Charly Poly put Jev next to open encoders, not GPT-5.6, and it lost on calibration

x.com

Charly Poly put Jev next to open encoders, not GPT-5.6, and it lost on calibration

x.com

A Hold'em table asks Jev about the opponent and lets code pick the bet

x.com

TechCrunch covers Jev, and quotes a Vercel engineer who swapped Luna for it

x.com

A CUA Speedrun chart prints Jev at 9.8% on OSWorld-50, 125 seconds a task

x.com

A native DuckDB extension batches Jev over SQL rows at about 2,300 rows per second on a live run

x.com

HiddenLayer prints Jev at 91% Confirm Price on a checkout ad, and this desk's jev-1.13.0 still chose place_order at 0.89

x.com

Opened October 16, Artificial Analysis Decision Index prints d1-ultra at 65 and Jev 1.14 at 57

x.com

On WANDS, more Jev questions move the mean to 0.610 and leave the median at 0.561

x.com

Stagehand, Chrome, and a11y trees: computer use as a Jev question

x.com

Opened October 19, JevBench ranks basal-omni-1.0-ultra first of 184 open systems on Capability at 82.5

x.com

Opened October 19, JevBench ranks basal-omni-1.0-ultra first of 184 open systems on Capability at 82.5

x.com

basal-1.5 scores 72.12% on Werdykt, and Jev 1.13.0 scores 81.64%

x.com

Opened October 19, JevBench ranks basal-omni-1.0-ultra first of 184 open systems on Capability at 82.5

x.com

Span-02's card prints Decision Index 0.3 text index 63.62, and the blog prints thinking off at 61.2 against Jev 58.0

x.com

Span-02's card prints Decision Index 0.3 text index 63.62, and the blog prints thinking off at 61.2 against Jev 58.0

x.com

Span-02's card prints Decision Index 0.3 text index 63.62, and the blog prints thinking off at 61.2 against Jev 58.0

x.com

Span-02's card prints Decision Index 0.3 text index 63.62, and the blog prints thinking off at 61.2 against Jev 58.0

x.com

Span-02's card prints Decision Index 0.3 text index 63.62, and the blog prints thinking off at 61.2 against Jev 58.0

x.com

Span-02's card prints Decision Index 0.3 text index 63.62, and the blog prints thinking off at 61.2 against Jev 58.0

x.com

Span-02's card prints Decision Index 0.3 text index 63.62, and the blog prints thinking off at 61.2 against Jev 58.0

x.com

toolgate 0.15 separates writing a script from running it, 19 of 20 on the paired set

x.com

OpenRouter, Cloudflare, and Netlify start serving Jev

x.com

jev-code's three-task ladder still fails after the call-nesting fixes

x.com

Rox's chart keeps 96% of relevance at 200 chunks, against 84% in production

x.com

One builder ran Jev on 1,500 of his own emails

x.com

GLiDE v1.2 prints Decision Index 0.4.1 Full 71.5, first of 229, at $0.04

x.com

GLiDE v1.2 prints Decision Index 0.4.1 Full 71.5, first of 229, at $0.04

x.com

Cloudflare posts Clef and Clef-flash, Apache 2.0 decision models on Workers AI GLiDE v1.2 prints Decision Index 0.4.1 Full 71.5, first of 229, at $0.04 Perplexity posted pplx-decider-v1.2 on October 16, with Decision Index 0.3 overall 68.09 on joint requests against v1.1 at 62.30

x.com

Sam Saffron tries Jev as a hidden control plane for GPT Live in term-llm

x.com

Jevrail scores a shell command before it runs, and the printed corpus leaves the model off

x.com

WaterSheep fine-tunes ModernBERT and serves /v1/systemone with a multi-label question

x.com

A 400M encoder named Kev scores 79.96% against Jev on 17,476 of its own rows

x.com

A 400M encoder named Kev scores 79.96% against Jev on 17,476 of its own rows

x.com

TypeSafe lists nine ways Jev 1.13 is still jagged, and asks for more

x.com

TypeSafe lists nine ways Jev 1.13 is still jagged, and asks for more

x.com

Microsoft-Decision-1 lists $0.042 per million input tokens and calls itself 4.5 times quicker than Quyet-1.0-Large

x.com

JevPilot is a playable driving sim that asks Jev which path to take

x.com

JevPilot is a playable driving sim that asks Jev which path to take

x.com

PolicyLM-1.7B scores each policy category from 0 to 1

x.com

jevmod scores sexual text at 0.984 AUROC, and ZeroNSFW keeps the pictures local

x.com

CLERC top-1 moves from 5% to 18% when jev-1.12 reranks the shortlist

x.com

One screenshot decision was $0.0002 on Jev and $0.032 on Opus 5

x.com

Opened October 17, JevBench API board v1.7.46 ranks kettei-fat first of 39 at composite 77.8

x.com

OpenBles Firewall 1 catches 119 of 120 agent attacks and flags no benign look-alike

x.com

OpenBles Firewall 1 catches 119 of 120 agent attacks and flags no benign look-alike

x.com

OpenBles Firewall 1 catches 119 of 120 agent attacks and flags no benign look-alike

x.com

StartLux-Decision v1.1 27B Instant prints Full score 69.12, first of 174 on Decision Index 0.3.2

x.com

StartLux-Decision v1.1 27B Instant prints Full score 69.12, first of 174 on Decision Index 0.3.2

x.com

StartLux-Decision v1.1 27B Instant prints Full score 69.12, first of 174 on Decision Index 0.3.2 Decision Index 0.2 prints Jev at 51.67, and a later post names another first place

x.com

TechCrunch covers Jev, and quotes a Vercel engineer who swapped Luna for it

x.com

Tetris as a legal-move test: Jev picks from the landings you give it

x.com

rev reads a frozen Qwen and posts 0.784 on public hard items

x.com

SubQ Snap 1.1 prints 65 on Artificial Analysis Decision Index, $0.04 per million input

x.com

SubQ Snap 1.1 prints 65 on Artificial Analysis Decision Index, $0.04 per million input

x.com

SubQ Snap 1.1 prints 65 on Artificial Analysis Decision Index, $0.04 per million input

x.com

On DeepEval's Paris weather example, jev-1.13.0 scored 0.254 when the answer invented a breeze

x.com

On DeepEval's Paris weather example, jev-1.13.0 scored 0.254 when the answer invented a breeze

x.com

Upstage's Solar Decide chart prints 87.0% on the public 231, and a p95 of 15.7 seconds

x.com

OpenAI Decisions API is generally available; this desk's October 23 POSTs returned HTTP 200

x.com

LangChain ran Jev as a judge: 500 matching pass/fail labels, and a $0.34 bill against Claude's $28

x.com

jym asks Jev which real subcommand a mistyped one was aiming at

x.com

A Japanese table has Jev at 96.0% on one set and 56.0% on JamC-QA

x.com

On October 26 TypeSafe's jev-latest named jev-1.14.0, and both aliases point at 1.14.0 OpenRouter, Cloudflare, and Netlify start serving Jev

x.com

Compaction without summarization: score the tool calls and drop them

x.com

AstroHan ran 30 coding tasks with a Jev context filter: 25 passed, 22 without it

x.com

TechCrunch covers Jev, and quotes a Vercel engineer who swapped Luna for it

x.com

TechCrunch covers Jev, and quotes a Vercel engineer who swapped Luna for it

x.com

TypeLLM said Jev 1.14 cannot return every vehicle box, including Tesla's flying Roadster; this desk's every-vehicle Box was one rectangle, n = 3, and type array returned HTTP 400

x.com

Glean replayed four internal decisions on Jev, and 40 expert transfers were 8.1 times faster

x.com

You.com's risk sweeps cost $0.14 each, and the report means tie

x.com

Arena scores typesafe/jev-router at +3.6% and $0.155 a task OpenRouter, Cloudflare, and Netlify start serving Jev

x.com

Arena scores typesafe/jev-router at +3.6% and $0.155 a task

x.com

SemIf, a 4B open rebuild that reads option logits, sits 0.7 behind Jev on JevBench

x.com

OpenAI Decisions API is generally available; this desk's October 23 POSTs returned HTTP 200

x.com

OpenAI Decisions API is generally available; this desk's October 23 POSTs returned HTTP 200

x.com

JevScout ranks oversized MCP text and prints 38% less agent time on its own check

x.com

Opened October 26, Decision Index 0.4.1 Full score for d3-turbo is 67.31, and vision full is 72.46

x.com

On eight Stripe passages, a 0-9 Score put Roll secret at 8.86 and the other six at 2.42 or below

x.com

Together prices Tev1 input at $0.04, and Ollama serves Choice, Score, and Noul

x.com

Celeris-1 Decision prints 80.0% on 3,210 jev-bench items at a 67 ms median, and Artificial Analysis lists it at 52

x.com

Celeris-1 Decision prints 80.0% on 3,210 jev-bench items at a 67 ms median, and Artificial Analysis lists it at 52

x.com

Glean replayed four internal decisions on Jev, and 40 expert transfers were 8.1 times faster

x.com

TypeLLM said Jev 1.14 cannot return every vehicle box, including Tesla's flying Roadster; this desk's every-vehicle Box was one rectangle, n = 3, and type array returned HTTP 400 TypeLLM scores Qwen3.8-27B at 228 of 231 public JevBench tasks when thinking is on

x.com

TypeLLM said Jev 1.14 cannot return every vehicle box, including Tesla's flying Roadster; this desk's every-vehicle Box was one rectangle, n = 3, and type array returned HTTP 400 TypeLLM scores Qwen3.8-27B at 228 of 231 public JevBench tasks when thinking is on

x.com

TypeLLM scores Qwen3.8-27B at 228 of 231 public JevBench tasks when thinking is on

x.com

TypeLLM said Jev 1.14 cannot return every vehicle box, including Tesla's flying Roadster; this desk's every-vehicle Box was one rectangle, n = 3, and type array returned HTTP 400

x.com

TypeLLM said Jev 1.14 cannot return every vehicle box, including Tesla's flying Roadster; this desk's every-vehicle Box was one rectangle, n = 3, and type array returned HTTP 400

x.com

TypeLLM scores Qwen3.8-27B at 228 of 231 public JevBench tasks when thinking is on

x.com

DSPy 3.4.0 calls Jev from Predict, with a gate at 0.2 and 0.6

x.com

You.com's risk sweeps cost $0.14 each, and the report means tie

x.com

Wrangle's chart prints 137.3 perfect matches per 200 with Jev

x.com

TypeSafe leaves stealth and ships Jev, a model that refuses to chat

x.com

TechCrunch covers Jev, and quotes a Vercel engineer who swapped Luna for it

x.com

OpenRouter, Cloudflare, and Netlify start serving Jev

x.com

OpenRouter, Cloudflare, and Netlify start serving Jev

x.com

Venice lists Jev on its API: same key, no copy of the state

x.com

TypeSafe lists nine ways Jev 1.13 is still jagged, and asks for more

x.com

TypeSafe drops the Jev waitlist; new accounts start with $5

x.com

TypeSafe-Danger-Zone-Long-Context raises both Jev context caps to 262k tokens

x.com

TypeSafe posted an API and console outage, then said it was back

x.com

TypeLLM said Jev 1.14 cannot return every vehicle box, including Tesla's flying Roadster; this desk's every-vehicle Box was one rectangle, n = 3, and type array returned HTTP 400

x.com

A green Continue button scored noul 0.99 on jev-1.14.0, n = 3, and an empty field scored 0.02

x.com

What Jev returns: System One in plain language

x.com

Deel's charts show 6 of 8 quality checks at parity, and two losses

x.com

On October 26 TypeSafe's jev-latest named jev-1.14.0, and both aliases point at 1.14.0

x.com

TypeSafe reopens Jev signups and drops the free credit for new accounts TypeSafe drops the Jev waitlist; new accounts start with $5

x.com

On October 26 TypeSafe's jev-latest named jev-1.14.0, and both aliases point at 1.14.0

x.com

TypeSafe reopens Jev signups and drops the free credit for new accounts

x.com

MotherDuck's prompt_jev() labels 100,000 AG News rows in 40 seconds for $0.50

x.com

A DTap chart puts a Jev 1.13 tool selector at 70.1% attack success, and a second Jev call at 36.1%

x.com

Arena scores typesafe/jev-router at +3.6% and $0.155 a task OpenRouter, Cloudflare, and Netlify start serving Jev

x.com

Rox's chart keeps 96% of relevance at 200 chunks, against 84% in production

x.com

TypeSafe reopens Jev signups and drops the free credit for new accounts

x.com

TypeSafe reopens Jev signups and drops the free credit for new accounts

x.com

TypeSafe reopens Jev signups and drops the free credit for new accounts

x.com

TypeSafe reopens Jev signups and drops the free credit for new accounts

x.com

Deel's charts show 6 of 8 quality checks at parity, and two losses

x.com

Deel's charts show 6 of 8 quality checks at parity, and two losses

x.com

Deel's charts show 6 of 8 quality checks at parity, and two losses

x.com

TypeSafe posted an API and console outage, then said it was back

x.com

On October 26 TypeSafe's jev-latest named jev-1.14.0, and both aliases point at 1.14.0

x.com

Open decision models start from an open language model

x.com

Opened October 16, Artificial Analysis Decision Index prints d1-ultra at 65 and Jev 1.14 at 57

x.com

Jev keeps the same direction on 28,074 classical-Korean sentences

x.com

OpenAI's Decisions API preview says results come back in 150 milliseconds

x.com

Venice lists Jev on its API: same key, no copy of the state

x.com

Venice lists Jev on its API: same key, no copy of the state

x.com

Vercel's October 2026 AI Gateway Production Index prints Jev at 35.3% of paid teams in its first two weeks

x.com

Opened October 16, Artificial Analysis Decision Index prints d1-ultra at 65 and Jev 1.14 at 57

x.com

Vercel's October 2026 AI Gateway Production Index prints Jev at 35.3% of paid teams in its first two weeks

x.com

Polylane moved five on-call decisions to Jev and prints a 39.1% lower bill per 1,000 calls

x.com

jev-gateway asks Jev which tool a coding agent should call

x.com

jev-gateway asks Jev which tool a coding agent should call

x.com

Violoop ran the same desktop harness with Jev and with Opus 5.5 on one QuickBooks bill

x.com

On eight Stripe passages, a 0-9 Score put Roll secret at 8.86 and the other six at 2.42 or below

x.com

Opened October 26, Decision Index 0.4.1 Full score for d3-ultra is 70.18, second of 229

x.com

Decision 2.0's Vega card prints 56.5 on its own Decision Index run

x.com

Vela 2.0's 9B card prints 41.63 on Decision Index 0.2.1

x.com

ToddlerJev prints 68.70% on MMAU-mini, with weights under the Qwen Research License

x.com

A green Continue button scored noul 0.99 on jev-1.14.0, n = 3, and an empty field scored 0.02

x.com

A green Continue button scored noul 0.99 on jev-1.14.0, n = 3, and an empty field scored 0.02

x.com

On 40 client requests, Jev and Opus both scored 37 of 40, at $0.0015 against $0.005 to $0.015

x.com

LangChain ships a Jev classifier, then parks it on the agent loop

x.com

Opened October 26, Decision Index 0.4.1 Full score for d3 is 67.15; d3-turbo and d3-ultra have their own pages

x.com

Opened October 26, Decision Index 0.4.1 Full score for d3 is 67.15; d3-turbo and d3-ultra have their own pages Decision 2.0's Vega card prints 56.5 on its own Decision Index run

x.com

Opened October 26, Decision Index 0.4.1 Full score for d3-turbo is 67.31, and vision full is 72.46

x.com

Opened October 26, Decision Index 0.4.1 Full score for d3-turbo is 67.31, and vision full is 72.46

x.com

Opened October 26, Decision Index 0.4.1 Full score for d3-ultra is 70.18, second of 229

x.com

Decision 2.0's Vega card prints 56.5 on its own Decision Index run

x.com

Vela 2.0's 9B card prints 41.63 on Decision Index 0.2.1

x.com

Opened October 26, Decision Index 0.4.1 Full score for d3-turbo is 67.31, and vision full is 72.46

x.com

On WANDS, more Jev questions move the mean to 0.610 and leave the median at 0.561

x.com

Bespoke Nimble 9B matches 90% of its own labels and 74.8% of a public suite

x.com

Bespoke Nimble 9B matches 90% of its own labels and 74.8% of a public suite

x.com

Decision Model Factory prints DF-JA-Large at 67.63 on S1MB Japanese Task Avg against a public snapshot high of 60.05

x.com

Decision Model Factory prints DF-JA-Large at 67.63 on S1MB Japanese Task Avg against a public snapshot high of 60.05

x.com

Decision Model Factory prints DF-JA-Large at 67.63 on S1MB Japanese Task Avg against a public snapshot high of 60.05

x.com

jevbench.app replays recorded Jev answers on labeled examples, and one test is 20 of 21

x.com

jevbench.app replays recorded Jev answers on labeled examples, and one test is 20 of 21

x.com

Zachi's pg-jev judges 2,000 Postgres rows in about 3.5 seconds

x.com

Zachi's pg-jev judges 2,000 Postgres rows in about 3.5 seconds

x.com

Zachi's pg-jev judges 2,000 Postgres rows in about 3.5 seconds

x.com

Zachi's pg-jev judges 2,000 Postgres rows in about 3.5 seconds

x.com

Open-Jev 27B v1.1 answers 197 of 231 public JevBench tasks

x.com

Open-Jev 27B v1.1 answers 197 of 231 public JevBench tasks

x.com

Open-Jev 27B v1.1 answers 197 of 231 public JevBench tasks

x.com

A DTap chart puts a Jev 1.13 tool selector at 70.1% attack success, and a second Jev call at 36.1%

x.com

A DTap chart puts a Jev 1.13 tool selector at 70.1% attack success, and a second Jev call at 36.1%

x.com

jev_local's Sarashina row beats LFM on two JGLUE tests, and the README says that is not a ranking

x.com

jevmod scores sexual text at 0.984 AUROC, and ZeroNSFW keeps the pictures local

x.com

One Jev pass scores 1,072 translations for $0.3165

x.com

Sthānika's Sieve-31B card prints Decision Index 0.3.2 Full score 63.8

x.com

Ollaya serves local decision models on port 11435, and prints winnow:e4b at 0.722

x.com

StartLux-Decision v1.1 27B Instant prints Full score 69.12, first of 174 on Decision Index 0.3.2

x.com

One post scores Jev at 100 and Laya's English checkpoint at 92, on 100 items

github.com

Jevsus ran a true/false diode on jev-1.13.0, and 45% of statements move with the wording

github.com

rev reads a frozen Qwen and posts 0.784 on public hard items

github.com

One Jev pass scores 1,072 translations for $0.3165

github.com

spark-s1-4b-v6 swaps in Qwen3.5 and prints a public-tier 83.1

github.com

On Ámbar Reflex the 4B logit player went further than the more accurate 9B

github.com

CLERC top-1 moves from 5% to 18% when jev-1.12 reranks the shortlist

github.com

CLERC top-1 moves from 5% to 18% when jev-1.12 reranks the shortlist

github.com

postgres-search can sink a bad hit, and the README has not measured that step

github.com

OR-decision-4B prints JevBench 193 of 231, and Choice 79.7 against Jev 79.6 on alphaXiv's chart

github.com

Decision Index 0.2 prints Jev at 51.67, and a later post names another first place

github.com

A 37-dataset paper scores frozen jev-1.13.0 against two open models

github.com

jev_local's Sarashina row beats LFM on two JGLUE tests, and the README says that is not a ranking

github.com

JEV Search moves NFCorpus nDCG@10 from 0.322 to 0.377

github.com

One screenshot decision was $0.0002 on Jev and $0.032 on Opus 5

github.com

jevelry turns a Jev call into a folder, a verdict, and a log

github.com

On WANDS, more Jev questions move the mean to 0.610 and leave the median at 0.561

github.com

Step-Jev scores 19.6% when the reward grades every search step

github.com

A 400M encoder named Kev scores 79.96% against Jev on 17,476 of its own rows

github.com

Bespoke Nimble 9B matches 90% of its own labels and 74.8% of a public suite

github.com

A week of Jev repos: routers, MCP, graphs, games

github.com

On WANDS, more Jev questions move the mean to 0.610 and leave the median at 0.561

github.com

Jevonian asks Jev which model should take the next coding turn

github.com

Bud Decision Studio serves eleven local decision models on port 8420

github.com

Eikos-27B prints 82.9 on its JevBench hard row, and 2.4% error once confidence is 0.90

github.com

Eikos-27B prints 82.9 on its JevBench hard row, and 2.4% error once confidence is 0.90

github.com

A DTap chart puts a Jev 1.13 tool selector at 70.1% attack success, and a second Jev call at 36.1%

github.com

mizchi's fit notes record 0 illegal moves in 489 MOBA calls, and 6 to 27 wrong pages where a one-line rule was 16/16

github.com

A Pi extension asks Jev which skills to load, and the 20-skill demo drops about 1,000 tokens

github.com

spark-s1-4b-v6 swaps in Qwen3.5 and prints a public-tier 83.1

github.com

Zachi's pg-jev judges 2,000 Postgres rows in about 3.5 seconds

github.com

On WANDS, more Jev questions move the mean to 0.610 and leave the median at 0.561

github.com

DSPy 3.4.0 calls Jev from Predict, with a gate at 0.2 and 0.6

github.com

A native DuckDB extension batches Jev over SQL rows at about 2,300 rows per second on a live run

github.com

On WANDS, more Jev questions move the mean to 0.610 and leave the median at 0.561

github.com

On DeepEval's Paris weather example, jev-1.13.0 scored 0.254 when the answer invented a breeze

github.com

CLM-8B caches the action list, and a fine-tuned head scores 31 of 38 DeepSWE holdouts

github.com

Plumb-4B reaches 89 of 111 public hard items, and the official board scored it 65.84

github.com

A CUA Speedrun chart prints Jev at 9.8% on OSWorld-50, 125 seconds a task

github.com

JevBench v1.5 ranks Cygnet 73.7 and Jev 1.13.0 72.1

github.com

swift-jev types Choice, Noul, and Score, and routes at 0.60 and 0.85

github.com

LangChain ran Jev as a judge: 500 matching pass/fail labels, and a $0.34 bill against Claude's $28

github.com

A 100,000-line needle takes 18.0 minutes with Jev and 24.7 without

github.com

djev reads DiffusionGemma as a decision API, and sits third on JevBench

github.com

llamacpp-jev answers /v1/systemone from an unmodified llama-server, including images

github.com

On WANDS, more Jev questions move the mean to 0.610 and leave the median at 0.561

github.com

mizchi's fit notes record 0 illegal moves in 489 MOBA calls, and 6 to 27 wrong pages where a one-line rule was 16/16

github.com

mizchi's fit notes record 0 illegal moves in 489 MOBA calls, and 6 to 27 wrong pages where a one-line rule was 16/16

github.com

A Pi extension asks Jev which skills to load, and the 20-skill demo drops about 1,000 tokens

github.com

fsdecide scores 91.3% on ForkScape review, and Jev scores 56.3%

github.com

DSPy 3.4.0 calls Jev from Predict, with a gate at 0.2 and 0.6

github.com

jevgrep solves the same 8 of 10 SWE-bench tasks, and the bill including Jev is 25.8% lower

github.com

A 100,000-line needle takes 18.0 minutes with Jev and 24.7 without

github.com

NIRNAY 450M prints 0.8792 on 3,080 Banking77 rows and says the Jev cell was not measured here

github.com

JevScout ranks oversized MCP text and prints 38% less agent time on its own check

github.com

Jevrail scores a shell command before it runs, and the printed corpus leaves the model off

github.com

Strands Decider 2B scores 167 of 231 public JevBench tasks

github.com

Strands Decider 2B scores 167 of 231 public JevBench tasks

github.com

OpenAI Decisions API is generally available; this desk's October 23 POSTs returned HTTP 200

github.com

On WANDS, more Jev questions move the mean to 0.610 and leave the median at 0.561

github.com

SimpleJev turns an open model into a classifier by reading the next-token logits

github.com

jevos-1b prints 0.815 on 2,000 yes/no questions, against Jev at 0.927

github.com

AgentOS posts a 121-turn routing table for an opt-in Jev strategy

github.com

Open-Jev 27B v1.1 answers 197 of 231 public JevBench tasks

github.com

fsdecide scores 91.3% on ForkScape review, and Jev scores 56.3%

github.com

Spring publishes a Java client that asks Jev before the chat model runs

github.com

On DeepEval's Paris weather example, jev-1.13.0 scored 0.254 when the answer invented a breeze

github.com

GitHub Next ships LocalJev, a prompted stand-in that speaks /v1/systemone

github.com

Step-Jev scores 19.6% when the reward grades every search step

github.com

DiffusionGemma as Jev reads one denoising step, and a hand chart sums to 198 of 201

github.com

A week of Jev repos: routers, MCP, graphs, games

github.com

ASO Workbench ranks App Store terms with a model trained on Jev's answers

github.com

Jev keeps the same direction on 28,074 classical-Korean sentences

github.com

dejevu books the same Flights search faster than the published Jev demo, on another machine

github.com

Jevis is a Flutter integration-test agent that only taps what you registered

github.com

Open decision models start from an open language model Jared Palmer's Kev family copies Jev's API on Qwen, and trails it on new sources

github.com

A buy-or-sell every Monad block, with a mock default model

github.com

Jev macOS Loop clicks native Mac windows; Finder sorted nine files in 7.39 seconds

github.com

LlamaIndex's Logan Markewich ships jeff, a local System One API on GLiFormer

github.com

DocJev classifies and splits real PDFs with Jev, 5.7x to 6.5x faster than Luna on the author's set

github.com

A week of Jev repos: routers, MCP, graphs, games

github.com

Jev keeps the same direction on 28,074 classical-Korean sentences

github.com

One Jev pass scores 1,072 translations for $0.3165

github.com

Step-Jev scores 19.6% when the reward grades every search step

github.com

A week of Jev repos: routers, MCP, graphs, games

github.com

A week of Jev repos: routers, MCP, graphs, games

github.com

A week of Jev repos: routers, MCP, graphs, games

github.com

dejevu books the same Flights search faster than the published Jev demo, on another machine A week of Jev repos: routers, MCP, graphs, games Browser Use wires Jev into the click loop: 7 seconds, $0.0039

github.com

A green Continue button scored noul 0.99 on jev-1.14.0, n = 3, and an empty field scored 0.02

github.com

imajev-4b: Hugging Face weights and JevBench 70.4 Plumb-4B reaches 89 of 111 public hard items, and the official board scored it 65.84 JevBench v1.5 ranks Cygnet 73.7 and Jev 1.13.0 72.1

github.com

TypeLLM scores Qwen3.8-27B at 228 of 231 public JevBench tasks when thinking is on

github.com

Open-Jev 27B v1.1 answers 197 of 231 public JevBench tasks

github.com

jevmod scores sexual text at 0.984 AUROC, and ZeroNSFW keeps the pictures local

github.com

jevmod scores sexual text at 0.984 AUROC, and ZeroNSFW keeps the pictures local

github.com

JevPilot is a playable driving sim that asks Jev which path to take

github.com

TypeLLM said Jev 1.14 cannot return every vehicle box, including Tesla's flying Roadster; this desk's every-vehicle Box was one rectangle, n = 3, and type array returned HTTP 400 A green Continue button scored noul 0.99 on jev-1.14.0, n = 3, and an empty field scored 0.02

github.com

A Hold'em table asks Jev about the opponent and lets code pick the bet

github.com

Stagehand, Chrome, and a11y trees: computer use as a Jev question

github.com

jsort ranks a file by asking Jev which of two texts is more of something

github.com

Open decision models start from an open language model

github.com

Open decision models start from an open language model

github.com

An n8n community node routes Jev answers onto category branches, including Needs Review

github.com

llamacpp-jev answers /v1/systemone from an unmodified llama-server, including images

github.com

Jevflake puts Jev in Snowflake: noul, choice, and score as SQL functions

github.com

jevsearch re-ranks keyword hits with Jev and scored 83% Hit@1 on the author's TypeSafe docs set

github.com

jgrep's test bench prints mean recall 0.926 across 60 commits, at $0.1112

github.com

Jeva-4B prints 0 flips on 600 Choice permutations, and BANKING77 at K=16 is 91.62 against Jev 82.50

github.com

llamacpp-jev answers /v1/systemone from an unmodified llama-server, including images

github.com

Cloudflare posts Clef and Clef-flash, Apache 2.0 decision models on Workers AI llamacpp-jev answers /v1/systemone from an unmodified llama-server, including images

github.com

llamacpp-jev answers /v1/systemone from an unmodified llama-server, including images Bespoke Nimble 9B matches 90% of its own labels and 74.8% of a public suite

github.com

llamacpp-jev answers /v1/systemone from an unmodified llama-server, including images

github.com

Cloudflare posts Clef and Clef-flash, Apache 2.0 decision models on Workers AI llamacpp-jev answers /v1/systemone from an unmodified llama-server, including images

github.com

Cloudflare posts Clef and Clef-flash, Apache 2.0 decision models on Workers AI llamacpp-jev answers /v1/systemone from an unmodified llama-server, including images Bespoke Nimble 9B matches 90% of its own labels and 74.8% of a public suite

github.com

llamacpp-jev answers /v1/systemone from an unmodified llama-server, including images

github.com

GitHub Next ships LocalJev, a prompted stand-in that speaks /v1/systemone

github.com

LlamaIndex's Logan Markewich ships jeff, a local System One API on GLiFormer

github.com

semgate asks Jev about each HTTP request, then a live range to attack it

github.com

semgate asks Jev about each HTTP request, then a live range to attack it

github.com

JevBench v1.5 ranks Cygnet 73.7 and Jev 1.13.0 72.1

github.com

jev-rubiks asks Jev when to speak, after a test where picking moves did nothing

github.com

Jev macOS Loop clicks native Mac windows; Finder sorted nine files in 7.39 seconds

github.com

pg-jev puts prompt_jev() in PostgreSQL and returns jsonb

github.com

TypeLLM scores Qwen3.8-27B at 228 of 231 public JevBench tasks when thinking is on

github.com

laya-mlx runs the Laya checkpoints on Apple Silicon

github.com

DiffusionGemma as Jev reads one denoising step, and a hand chart sums to 198 of 201

github.com

jevmod scores sexual text at 0.984 AUROC, and ZeroNSFW keeps the pictures local

github.com

imajev-4b: Hugging Face weights and JevBench 70.4

github.com

hotdog-27B leads a 1,631-decision holdout and trails on 74 public items

github.com

ASO Workbench ranks App Store terms with a model trained on Jev's answers

github.com

Nace posts Drex 1.1, and the card titles the weights Drex DLM

github.com

Nace posts Drex 1.1, and the card titles the weights Drex DLM

github.com

llamacpp-jev answers /v1/systemone from an unmodified llama-server, including images

github.com

Laya, an Apache 2.0 System One stack, publishes where it beats Jev's numbers and where it does not

github.com

Vega, the vegaml package, scores 75.4% on 800 phishing emails against Jev at 61.9%

github.com

jevctl puts Jev on the command line, with a compaction hook that keeps lines verbatim

github.com

Quyet-1.0-Large's homepage prints JevBench v1.6.0 at 71.39

github.com

BranchOne returns Choice probabilities from Qwen3.5-0.8B-Base, median 0.305 s on ten short text questions

github.com

A week of Jev repos: routers, MCP, graphs, games

github.com

Jev scored 384 morning headlines in 25 seconds; Opus 5 finished four

github.com

Jev scored 384 morning headlines in 25 seconds; Opus 5 finished four

github.com

TypeLLM scores Qwen3.8-27B at 228 of 231 public JevBench tasks when thinking is on

github.com

jevmod scores sexual text at 0.984 AUROC, and ZeroNSFW keeps the pictures local

github.com

Ollaya serves local decision models on port 11435, and prints winnow:e4b at 0.722

github.com

A week of Jev repos: routers, MCP, graphs, games

github.com

Aman Kumar ran about 16,000 Jev calls and called it a filter, not a classifier

github.com

OpenAI Decisions API is generally available; this desk's October 23 POSTs returned HTTP 200

github.com

OpenAI Decisions API is generally available; this desk's October 23 POSTs returned HTTP 200

github.com

astra-jev-harness cuts Astra input 27% on one two-file fixture

github.com

Step-Jev scores 19.6% when the reward grades every search step

github.com

The Jeeves README prints 0.865 on the public JevBench hard tier

github.com

A native DuckDB extension batches Jev over SQL rows at about 2,300 rows per second on a live run

github.com

A week of Jev repos: routers, MCP, graphs, games

github.com

Bespoke Nimble 9B matches 90% of its own labels and 74.8% of a public suite

github.com

Upstage's Solar Decide chart prints 87.0% on the public 231, and a p95 of 15.7 seconds JevBench v1.5 ranks Cygnet 73.7 and Jev 1.13.0 72.1

github.com

On eight Stripe passages, a 0-9 Score put Roll secret at 8.86 and the other six at 2.42 or below

github.com

OpenBles Firewall 1 catches 119 of 120 agent attacks and flags no benign look-alike

github.com

Zachi's pg-jev judges 2,000 Postgres rows in about 3.5 seconds

github.com

On WANDS, more Jev questions move the mean to 0.610 and leave the median at 0.561

github.com

On WANDS, more Jev questions move the mean to 0.610 and leave the median at 0.561

github.com

jev-code's three-task ladder still fails after the call-nesting fixes

github.com

toolgate 0.15 separates writing a script from running it, 19 of 20 on the paired set

github.com

Rizzo Flow is 49 ms on an RTX 5060 Ti, and six missing-evidence rows stay confident

github.com

Opened October 19, JevBench ranks basal-omni-1.0-ultra first of 184 open systems on Capability at 82.5 basal-1.5 scores 72.12% on Werdykt, and Jev 1.13.0 scores 81.64%

github.com

A 100-case security triage demo put Jev at 65.3% balanced accuracy against Terra 75.1% and Opus 91.2%

github.com

fsdecide scores 91.3% on ForkScape review, and Jev scores 56.3%

github.com

A week of Jev repos: routers, MCP, graphs, games

github.com

JevBench ranks Sage 1.3.0 first of 27 hosted systems at composite 74.0

github.com

Jevrail scores a shell command before it runs, and the printed corpus leaves the model off

github.com

WaterSheep fine-tunes ModernBERT and serves /v1/systemone with a multi-label question

github.com

jev_local's Sarashina row beats LFM on two JGLUE tests, and the README says that is not a ranking

github.com

Opened October 26, Decision Index 0.4.1 Full score for d3-ultra is 70.18, second of 229 Opened October 26, Decision Index 0.4.1 Full score for d3-turbo is 67.31, and vision full is 72.46 Opened October 26, Decision Index 0.4.1 Full score for d3 is 67.15; d3-turbo and d3-ultra have their own pages

github.com

A week of Jev repos: routers, MCP, graphs, games

github.com

Rizzo Flow is 49 ms on an RTX 5060 Ti, and six missing-evidence rows stay confident SemIf, a 4B open rebuild that reads option logits, sits 0.7 behind Jev on JevBench

github.com

llamacpp-jev answers /v1/systemone from an unmodified llama-server, including images

github.com

On 40 client requests, Jev and Opus both scored 37 of 40, at $0.0015 against $0.005 to $0.015

github.com

A 400M encoder named Kev scores 79.96% against Jev on 17,476 of its own rows

github.com

Spring publishes a Java client that asks Jev before the chat model runs

github.com

SREGym put Jev on a Codex agent: 24 passes out of 50, up from 20

github.com

SREGym put Jev on a Codex agent: 24 passes out of 50, up from 20

github.com

DSPy 3.4.0 calls Jev from Predict, with a gate at 0.2 and 0.6

github.com

StartLux-Decision v1.1 27B Instant prints Full score 69.12, first of 174 on Decision Index 0.3.2 Decision Index 0.2 prints Jev at 51.67, and a later post names another first place

github.com

OpenBles Firewall 1 catches 119 of 120 agent attacks and flags no benign look-alike

github.com

Step-Jev scores 19.6% when the reward grades every search step

github.com

Step-Jev scores 19.6% when the reward grades every search step

github.com

Sthānika's Sieve-31B card prints Decision Index 0.3.2 Full score 63.8

github.com

Strands Decider 2B scores 167 of 231 public JevBench tasks

github.com

On WANDS, more Jev questions move the mean to 0.610 and leave the median at 0.561

github.com

RLCDAlignBench puts a generic Noul at median AUROC 0.886, and median ECE at 0.168

github.com

jym asks Jev which real subcommand a mistyped one was aiming at

github.com

A Cloud Run recipe serves DiffusionGemma-Jev at 35 to 60 ms

github.com

Compaction without summarization: score the tool calls and drop them

github.com

JevGate turns one function at a time into a finding at 0.80

github.com

Sam Saffron tries Jev as a hidden control plane for GPT Live in term-llm

github.com

jgrep's test bench prints mean recall 0.926 across 60 commits, at $0.1112

github.com

Five runs leave a 3.3-point lead on 18 claim checks

github.com

postgres-search can sink a bad hit, and the README has not measured that step

github.com

Five runs leave a 3.3-point lead on 18 claim checks

github.com

TypeLLM scores Qwen3.8-27B at 228 of 231 public JevBench tasks when thinking is on

github.com

Open decision models start from an open language model Together prices Tev1 input at $0.04, and Ollama serves Choice, Score, and Noul

github.com

jevgrep solves the same 8 of 10 SWE-bench tasks, and the bill including Jev is 25.8% lower

github.com

One Jev pass scores 1,072 translations for $0.3165

github.com

On 40 client requests, Jev and Opus both scored 37 of 40, at $0.0015 against $0.005 to $0.015

github.com

Bespoke Nimble 9B matches 90% of its own labels and 74.8% of a public suite

github.com

TypeLLM scores Qwen3.8-27B at 228 of 231 public JevBench tasks when thinking is on

github.com

Hono's author routes HTTP by meaning, then warns you not to

github.com

A week of Jev repos: routers, MCP, graphs, games

github.com

A week of Jev repos: routers, MCP, graphs, games

github.com

AgentOS posts a 121-turn routing table for an opt-in Jev strategy

github.com

PageIndex walks long filings with Jev Choice calls and checks two answers

github.com

PageIndex walks long filings with Jev Choice calls and checks two answers

github.com

Vega, the vegaml package, scores 75.4% on 800 phishing emails against Jev at 61.9%

github.com

Vercel's form router keeps Jev's destination when confidence is at least 0.95

github.com

Bespoke Nimble 9B matches 90% of its own labels and 74.8% of a public suite

github.com

jev-gateway asks Jev which tool a coding agent should call

github.com

jev-gateway asks Jev which tool a coding agent should call

github.com

DiffusionGemma as Jev reads one denoising step, and a hand chart sums to 198 of 201

github.com

DiffusionGemma as Jev reads one denoising step, and a hand chart sums to 198 of 201

github.com

A week of Jev repos: routers, MCP, graphs, games Compaction without summarization: score the tool calls and drop them

github.com

Jevonian asks Jev which model should take the next coding turn

github.com

JevScout ranks oversized MCP text and prints 38% less agent time on its own check

github.com

You.com's risk sweeps cost $0.14 each, and the report means tie

github.com

Hono's author routes HTTP by meaning, then warns you not to

github.com

MiniCPM5-2B-Jev prints 78.79% on JevBench, first among open models at or under 2B

github.com

DiffusionGemma as Jev reads one denoising step, and a hand chart sums to 198 of 201

github.com

Open-Jev 27B v1.1 answers 197 of 231 public JevBench tasks

github.com

jevmod scores sexual text at 0.984 AUROC, and ZeroNSFW keeps the pictures local

github.com

jevmod scores sexual text at 0.984 AUROC, and ZeroNSFW keeps the pictures local

github.com

jevmod scores sexual text at 0.984 AUROC, and ZeroNSFW keeps the pictures local

github.com

A frozen Jev pipeline reports the lowest ten-corpus valence-arousal error in a SemEval table

ai-sdk.dev

TypeSafe starts pulling people off the Jev waitlist

ai.azure.com

Microsoft-Decision-1 lists $0.042 per million input tokens and calls itself 4.5 times quicker than Quyet-1.0-Large

ai.google.dev

DiffusionGemma as Jev reads one denoising step, and a hand chart sums to 198 of 201

aibenchy.com

On October 26 TypeSafe's jev-latest named jev-1.14.0, and both aliases point at 1.14.0

amankumar.ai

Aman Kumar ran about 16,000 Jev calls and called it a filter, not a classifier

api.hal-x.ai

THX-01 averages 98.4 on its 2,843 support tickets

artificialanalysis.ai

Opened October 16, Artificial Analysis Decision Index prints d1-ultra at 65 and Jev 1.14 at 57 Celeris-1 Decision prints 80.0% on 3,210 jev-bench items at a 67 ms median, and Artificial Analysis lists it at 52

artificialanalysis.ai

Opened October 16, Artificial Analysis Decision Index prints d1-ultra at 65 and Jev 1.14 at 57 Celeris-1 Decision prints 80.0% on 3,210 jev-bench items at a 67 ms median, and Artificial Analysis lists it at 52

arxiv.org

Jeva-4B prints 0 flips on 600 Choice permutations, and BANKING77 at K=16 is 91.62 against Jev 82.50

arxiv.org

Qwen-Audio-Jev prints 78.70% on MMAU-mini and about 20x the base model's speed

arxiv.org

Decision Model Factory prints DF-JA-Large at 67.63 on S1MB Japanese Task Avg against a public snapshot high of 60.05

arxiv.org

Decision Model Factory prints DF-JA-Large at 67.63 on S1MB Japanese Task Avg against a public snapshot high of 60.05

arxiv.org

A 37-dataset paper scores frozen jev-1.13.0 against two open models

arxiv.org

A v3 judge paper puts a frozen Jev cascade 0.9 points above GPT-6 on 1,610 pairs

arxiv.org

A registered attack on jev-1.13.0 moves validated success from 1.8% to 3.5%

arxiv.org

RLCDAlignBench puts a generic Noul at median AUROC 0.886, and median ECE at 0.168

arxiv.org

JevAdvBench's abstract says one appended opinion flips 12.1% of jev-1.13.0 decisions

arxiv.org

A frozen Jev pipeline reports the lowest ten-corpus valence-arousal error in a SemEval table

arxiv.org

A 37-dataset paper scores frozen jev-1.13.0 against two open models

arxiv.org

StreamDecisionBench puts Kev-27B at 62.15 and Jev at 59.63

arxiv.org

HydroJEV tests jev-1.13.0 as a one-second screen on simulated water-network alarms

arxiv.org

A DTap chart puts a Jev 1.13 tool selector at 70.1% attack success, and a second Jev call at 36.1%

arxiv.org

A v3 judge paper puts a frozen Jev cascade 0.9 points above GPT-6 on 1,610 pairs

arxiv.org

A registered attack on jev-1.13.0 moves validated success from 1.8% to 3.5%

arxiv.org

Qwen-Audio-Jev prints 78.70% on MMAU-mini and about 20x the base model's speed

arxiv.org

HydroJEV tests jev-1.13.0 as a one-second screen on simulated water-network alarms

arxiv.org

A frozen Jev pipeline reports the lowest ten-corpus valence-arousal error in a SemEval table

arxiv.org

A 37-dataset paper scores frozen jev-1.13.0 against two open models

azure.microsoft.com

Microsoft-Decision-1 lists $0.042 per million input tokens and calls itself 4.5 times quicker than Quyet-1.0-Large

basal.si5.pl

basal-1.5 scores 72.12% on Werdykt, and Jev 1.13.0 scores 81.64%

basal.si5.pl

basal-1.5 scores 72.12% on Werdykt, and Jev 1.13.0 scores 81.64%

basal.si5.pl

Opened October 19, JevBench ranks basal-omni-1.0-ultra first of 184 open systems on Capability at 82.5

basal.si5.pl

Opened October 19, JevBench ranks basal-omni-1.0-ultra first of 184 open systems on Capability at 82.5

basal.si5.pl

basal-1.5 scores 72.12% on Werdykt, and Jev 1.13.0 scores 81.64%

basal.si5.pl

basal-1.5 scores 72.12% on Werdykt, and Jev 1.13.0 scores 81.64%

baseten.co

Base Labs posts mammoth-1 at Decision Index 0.4 Full 71.24; live 0.4.1 prints 68.4

baseten.co

Base Labs posts mammoth-1 at Decision Index 0.4 Full 71.24; live 0.4.1 prints 68.4

baseten.co

Base Labs posts mammoth-1 at Decision Index 0.4 Full 71.24; live 0.4.1 prints 68.4

benchmarkheaven.com

JevBench v1.5 ranks Cygnet 73.7 and Jev 1.13.0 72.1

benchmarkheaven.com

Quyet-1.0-Large's homepage prints JevBench v1.6.0 at 71.39

benchmarkheaven.com

Opened October 17, JevBench API board v1.7.46 ranks kettei-fat first of 39 at composite 77.8 imajev-4b: Hugging Face weights and JevBench 70.4

benchmarkheaven.com

imajev-4b: Hugging Face weights and JevBench 70.4

benchmarkheaven.com

Opened October 17, JevBench API board v1.7.46 ranks kettei-fat first of 39 at composite 77.8 JevBench ranks Sage 1.3.0 first of 27 hosted systems at composite 74.0 Microsoft-Decision-1 lists $0.042 per million input tokens and calls itself 4.5 times quicker than Quyet-1.0-Large OpenAI Decisions API is generally available; this desk's October 23 POSTs returned HTTP 200 GLiDE v1.2 prints Decision Index 0.4.1 Full 71.5, first of 229, at $0.04 Liquid's chart puts d1 at 58.9 on an internal Decision Index 0.2.1 JevBench v1.5 ranks Cygnet 73.7 and Jev 1.13.0 72.1

benchmarkheaven.com

Opened October 19, JevBench ranks basal-omni-1.0-ultra first of 184 open systems on Capability at 82.5 H2O-Lightning-4B v1.1 prints JevBench composite 72.5 against Jev at 71.5 Nace posts Drex 1.5 at $0.04, and its chart shows 199 of 231 against Jev's 201 Inception calls Mercury Decide first on OpenRouter's JevBench v1.4, and prints no score imajev-4b: Hugging Face weights and JevBench 70.4 djev reads DiffusionGemma as a decision API, and sits third on JevBench SimpleJev turns an open model into a classifier by reading the next-token logits Jared Palmer's Kev family copies Jev's API on Qwen, and trails it on new sources Laya, an Apache 2.0 System One stack, publishes where it beats Jev's numbers and where it does not JevBench v1.5 ranks Cygnet 73.7 and Jev 1.13.0 72.1 SemIf, a 4B open rebuild that reads option logits, sits 0.7 behind Jev on JevBench

benchmarkheaven.com

Opened October 19, JevBench ranks basal-omni-1.0-ultra first of 184 open systems on Capability at 82.5 On October 26 TypeSafe's jev-latest named jev-1.14.0, and both aliases point at 1.14.0

benchmarkheaven.com

JevBench v1.5 ranks Cygnet 73.7 and Jev 1.13.0 72.1

blog.cloudflare.com

Clef-omni scores a 21-second video in about 1.5 seconds, at $0.15 per million input tokens Cloudflare posts Clef and Clef-flash, Apache 2.0 decision models on Workers AI

blog.cloudflare.com

Cloudflare posts Clairic, a System One extraction model at $0.10 per million input

blog.cloudflare.com

hunch runs EmbeddingGemma2 as a CPU System One endpoint at 20 ms on text and 100 ms on images Open decision models start from an open language model Cloudflare posts Clef and Clef-flash, Apache 2.0 decision models on Workers AI

blog.dailydoseofds.com

CLERC top-1 moves from 5% to 18% when jev-1.12 reranks the shortlist

blog.google

hunch runs EmbeddingGemma2 as a CPU System One endpoint at 20 ms on text and 100 ms on images

celeris.ai

Celeris-1 Decision prints 80.0% on 3,210 jev-bench items at a 67 ms median, and Artificial Analysis lists it at 52

claude.ai

A chart puts Jev's reading of 25 probability words next to 99 people

commandline.microsoft.com

Microsoft-Decision-1 lists $0.042 per million input tokens and calls itself 4.5 times quicker than Quyet-1.0-Large

completeskeptic.com

Almeida's bet on Jevons paradox over chatbot price cuts

composio.dev

On 40 client requests, Jev and Opus both scored 37 of 40, at $0.0015 against $0.005 to $0.015

console.nace.ai

Nace posts Drex Max at Decision Index 0.3.2 public 67.59, $0.06 per million input

console.sqwish.ai

Opened October 17, JevBench API board v1.7.46 ranks kettei-fat first of 39 at composite 77.8

console.subq.ai

SubQ Snap 1.1 prints 65 on Artificial Analysis Decision Index, $0.04 per million input

contrastive-lm.notion.site

CLM-8B caches the action list, and a fine-tuned head scores 31 of 38 DeepSWE holdouts

crates.io

JevGate turns one function at a time into a finding at 0.80

cuaspeedrun.com

A CUA Speedrun chart prints Jev at 9.8% on OSWorld-50, 125 seconds a task

databricks.com

Databricks posts ai_decide, a Beta function that returns a noul, a choice, or a score

decision.vllm-sr.ai

Opened October 26, Decision Index 0.4.1 Full score for d3-ultra is 70.18, second of 229 Opened October 26, Decision Index 0.4.1 Full score for d3-turbo is 67.31, and vision full is 72.46

decodingtrust-agent.com

A DTap chart puts a Jev 1.13 tool selector at 70.1% attack success, and a second Jev call at 36.1%

deepeval.com

On DeepEval's Paris weather example, jev-1.13.0 scored 0.254 when the answer invented a breeze

deepeval.com

On DeepEval's Paris weather example, jev-1.13.0 scored 0.254 when the answer invented a breeze

deepeval.com

On DeepEval's Paris weather example, jev-1.13.0 scored 0.254 when the answer invented a breeze

deepeval.com

On DeepEval's Paris weather example, jev-1.13.0 scored 0.254 when the answer invented a breeze

deepeval.com

On DeepEval's Paris weather example, jev-1.13.0 scored 0.254 when the answer invented a breeze

demos.exa.ai

Exa posts a demo that puts web results into Jev's state

developers.cloudflare.com

Cloudflare posts Clef and Clef-flash, Apache 2.0 decision models on Workers AI

developers.cloudflare.com

Hono's author routes HTTP by meaning, then warns you not to OpenRouter, Cloudflare, and Netlify start serving Jev

developers.cloudflare.com

On October 26 TypeSafe's jev-latest named jev-1.14.0, and both aliases point at 1.14.0 OpenRouter, Cloudflare, and Netlify start serving Jev

developers.cloudflare.com

Cloudflare posts Clairic, a System One extraction model at $0.10 per million input

developers.cloudflare.com

Cloudflare posts Clef and Clef-flash, Apache 2.0 decision models on Workers AI

developers.cloudflare.com

Cloudflare posts Clef and Clef-flash, Apache 2.0 decision models on Workers AI

developers.cloudflare.com

Clef-omni scores a 21-second video in about 1.5 seconds, at $0.15 per million input tokens

developers.openai.com

OpenAI Decisions API is generally available; this desk's October 23 POSTs returned HTTP 200

developers.openai.com

OpenAI Decisions API is generally available; this desk's October 23 POSTs returned HTTP 200 OpenAI's Decisions API preview says results come back in 150 milliseconds

developers.openai.com

OpenAI Decisions API is generally available; this desk's October 23 POSTs returned HTTP 200

developers.openai.com

OpenAI Decisions API is generally available; this desk's October 23 POSTs returned HTTP 200

djev.dev

djev reads DiffusionGemma as a decision API, and sits third on JevBench

djev.dev

djev reads DiffusionGemma as a decision API, and sits third on JevBench

docs.celeris.ai

Celeris-1 Decision prints 80.0% on 3,210 jev-bench items at a 67 ms median, and Artificial Analysis lists it at 52

docs.celeris.ai

Celeris-1 Decision prints 80.0% on 3,210 jev-bench items at a 67 ms median, and Artificial Analysis lists it at 52

docs.celeris.ai

Celeris-1 Decision prints 80.0% on 3,210 jev-bench items at a 67 ms median, and Artificial Analysis lists it at 52

docs.celeris.ai

Celeris-1 Decision prints 80.0% on 3,210 jev-bench items at a 67 ms median, and Artificial Analysis lists it at 52

docs.databricks.com

Databricks posts ai_decide, a Beta function that returns a noul, a choice, or a score

docs.databricks.com

Databricks posts ai_decide, a Beta function that returns a noul, a choice, or a score

docs.fastino.ai

GLiDE v1.2 prints Decision Index 0.4.1 Full 71.5, first of 229, at $0.04

docs.fastino.ai

GLiDE v1.2 prints Decision Index 0.4.1 Full 71.5, first of 229, at $0.04

docs.langchain.com

LangChain ships a Jev classifier, then parks it on the agent loop

docs.langchain.com

Datadog prints Jev Noul recall 82.90% to 3.88% under a counterfeit policy, and this desk's is_risky fell from 0.94 to 0.13

docs.liquid.ai

Open d1's d1-3B card prints 48.57 on a self-run of Decision Index 0.2.1

docs.liquid.ai

Opened October 16, Artificial Analysis Decision Index prints d1-ultra at 65 and Jev 1.14 at 57 Open d1's d1-3B card prints 48.57 on a self-run of Decision Index 0.2.1 Liquid's chart puts d1 at 58.9 on an internal Decision Index 0.2.1

docs.litellm.ai

On October 26 TypeSafe's jev-latest named jev-1.14.0, and both aliases point at 1.14.0 TypeLLM said Jev 1.14 cannot return every vehicle box, including Tesla's flying Roadster; this desk's every-vehicle Box was one rectangle, n = 3, and type array returned HTTP 400 A green Continue button scored noul 0.99 on jev-1.14.0, n = 3, and an empty field scored 0.02 TypeSafe-Danger-Zone-Long-Context raises both Jev context caps to 262k tokens LiteLLM adds day-0 Jev 1.14 images, box questions, and long context on both decision routes

docs.litellm.ai

TypeLLM said Jev 1.14 cannot return every vehicle box, including Tesla's flying Roadster; this desk's every-vehicle Box was one rectangle, n = 3, and type array returned HTTP 400 A green Continue button scored noul 0.99 on jev-1.14.0, n = 3, and an empty field scored 0.02 TypeSafe-Danger-Zone-Long-Context raises both Jev context caps to 262k tokens LiteLLM adds day-0 Jev 1.14 images, box questions, and long context on both decision routes

docs.litellm.ai

LiteLLM adds day-0 Jev 1.14 images, box questions, and long context on both decision routes

docs.litellm.ai

LiteLLM adds day-0 Jev 1.14 images, box questions, and long context on both decision routes

docs.litellm.ai

LiteLLM adds day-0 Jev 1.14 images, box questions, and long context on both decision routes

docs.perplexity.ai

Perplexity posted pplx-decider-v1.2 on October 16, with Decision Index 0.3 overall 68.09 on joint requests against v1.1 at 62.30

docs.snowflake.com

Abe's 1,774 Snowflake posts cost $0.055 on Jev Noul against $1.87 on AI_FILTER

docs.sqwish.ai

Opened October 17, JevBench API board v1.7.46 ranks kettei-fat first of 39 at composite 77.8

docs.sqwish.ai

Opened October 17, JevBench API board v1.7.46 ranks kettei-fat first of 39 at composite 77.8

docs.sqwish.ai

Opened October 17, JevBench API board v1.7.46 ranks kettei-fat first of 39 at composite 77.8

docs.subq.ai

SubQ Snap 1.1 prints 65 on Artificial Analysis Decision Index, $0.04 per million input

docs.subq.ai

SubQ Snap 1.1 prints 65 on Artificial Analysis Decision Index, $0.04 per million input

docs.subq.ai

SubQ Snap 1.1 prints 65 on Artificial Analysis Decision Index, $0.04 per million input

docs.subq.ai

SubQ Snap 1.1 prints 65 on Artificial Analysis Decision Index, $0.04 per million input

docs.subq.ai

SubQ Snap 1.1 prints 65 on Artificial Analysis Decision Index, $0.04 per million input

doi.org

A 37-dataset paper scores frozen jev-1.13.0 against two open models

eikos-chess.vercel.app

Eikos-27B prints 82.9 on its JevBench hard row, and 2.4% error once confidence is 0.90

every.to

Every ran 777 judgments in under a second and still wanted a better accuracy check

fastino.ai

Open decision models start from an open language model

fastino.ai

GLiDE v1.2 prints Decision Index 0.4.1 Full 71.5, first of 229, at $0.04 Decision Index 0.2 prints Jev at 51.67, and a later post names another first place

fastino.ai

GLiDE v1.2 prints Decision Index 0.4.1 Full 71.5, first of 229, at $0.04

ft.com

TechCrunch covers Jev, and quotes a Vercel engineer who swapped Luna for it

gist.github.com

Lightpanda's hostel script counts browser memory, and the Jev clock stays in the post

h2o.ai

H2O-Lightning-4B v1.1 prints JevBench composite 72.5 against Jev at 71.5

hackersquad.io

TechCrunch prices one Jev monitor at $2.94, and the demo page prints another estimate

hails.info

A chart puts Jev's reading of 25 probability words next to 99 people

hevmind.com

On WANDS, more Jev questions move the mean to 0.610 and leave the median at 0.561

hiddenlayer.com

HiddenLayer prints Jev at 91% Confirm Price on a checkout ad, and this desk's jev-1.13.0 still chose place_order at 0.89

hiddenlayer.com

HiddenLayer prints Jev at 91% Confirm Price on a checkout ad, and this desk's jev-1.13.0 still chose place_order at 0.89

hindsight.vectorize.io

CLERC top-1 moves from 5% to 18% when jev-1.12 reranks the shortlist

hindsight.vectorize.io

CLERC top-1 moves from 5% to 18% when jev-1.12 reranks the shortlist

huggingface.co

laya-mlx runs the Laya checkpoints on Apple Silicon

huggingface.co

PurpleMIST-Flash-1.0 scores 0.668 on 15,050 translated African-language decisions

huggingface.co

OR-decision-4B prints JevBench 193 of 231, and Choice 79.7 against Jev 79.6 on alphaXiv's chart

huggingface.co

Eikos-27B prints 82.9 on its JevBench hard row, and 2.4% error once confidence is 0.90

huggingface.co

AutoTrust scores GEV-26B-Decide at 62.48 on its own Decision Index run Decision Index 0.2 prints Jev at 51.67, and a later post names another first place

huggingface.co

Bespoke Nimble 9B matches 90% of its own labels and 74.8% of a public suite

huggingface.co

llamacpp-jev answers /v1/systemone from an unmodified llama-server, including images Bespoke Nimble 9B matches 90% of its own labels and 74.8% of a public suite

huggingface.co

Blink v0.3 prints Decision Index 0.2.1 at 57.48

huggingface.co

Eikos-27B prints 82.9 on its JevBench hard row, and 2.4% error once confidence is 0.90

huggingface.co

Quyet-1.0-Large's homepage prints JevBench v1.6.0 at 71.39

huggingface.co

Quyet-1.0-Large's homepage prints JevBench v1.6.0 at 71.39

huggingface.co

Cloudflare posts Clef and Clef-flash, Apache 2.0 decision models on Workers AI

huggingface.co

Cloudflare posts Clef and Clef-flash, Apache 2.0 decision models on Workers AI

huggingface.co

Clef-omni scores a 21-second video in about 1.5 seconds, at $0.15 per million input tokens

huggingface.co

CLM-8B caches the action list, and a fine-tuned head scores 31 of 38 DeepSWE holdouts

huggingface.co

Laya, an Apache 2.0 System One stack, publishes where it beats Jev's numbers and where it does not

huggingface.co

Open d1's d1-3B card prints 48.57 on a self-run of Decision Index 0.2.1

huggingface.co

Decision 2.0's Vega card prints 56.5 on its own Decision Index run

huggingface.co

Opened October 26, Decision Index 0.4.1 Full score for d3-ultra is 70.18, second of 229 Opened October 26, Decision Index 0.4.1 Full score for d3-turbo is 67.31, and vision full is 72.46 Opened October 26, Decision Index 0.4.1 Full score for d3 is 67.15; d3-turbo and d3-ultra have their own pages

huggingface.co

Opened October 26, Decision Index 0.4.1 Full score for d3-ultra is 70.18, second of 229 Base Labs posts mammoth-1 at Decision Index 0.4 Full 71.24; live 0.4.1 prints 68.4 Opened October 26, Decision Index 0.4.1 Full score for d3-turbo is 67.31, and vision full is 72.46 Span-02's card prints Decision Index 0.3 text index 63.62, and the blog prints thinking off at 61.2 against Jev 58.0 SubQ Snap 1.1 prints 65 on Artificial Analysis Decision Index, $0.04 per million input Nace posts Drex 1.1, and the card titles the weights Drex DLM GLiDE v1.2 prints Decision Index 0.4.1 Full 71.5, first of 229, at $0.04 Perplexity posted pplx-decider-v1.2 on October 16, with Decision Index 0.3 overall 68.09 on joint requests against v1.1 at 62.30 Decision Index 0.2 prints Jev at 51.67, and a later post names another first place

huggingface.co

Decision 2.0's Vega card prints 56.5 on its own Decision Index run

huggingface.co

Decision 2.0's Vega card prints 56.5 on its own Decision Index run

huggingface.co

Decision 2.0's Vega card prints 56.5 on its own Decision Index run

huggingface.co

Decision 2.0's Vega card prints 56.5 on its own Decision Index run

huggingface.co

Decision 2.0's Vega card prints 56.5 on its own Decision Index run

huggingface.co

Decision 2.0's Vega card prints 56.5 on its own Decision Index run Decision Index 0.2 prints Jev at 51.67, and a later post names another first place

huggingface.co

Jevre's test split is 0.9323, and held-out sources fall to 0.8366

huggingface.co

THX-01 averages 98.4 on its 2,843 support tickets

huggingface.co

Nace posts Drex 1.5 at $0.04, and its chart shows 199 of 231 against Jev's 201

huggingface.co

Eikos-27B prints 82.9 on its JevBench hard row, and 2.4% error once confidence is 0.90

huggingface.co

NIRNAY 450M prints 0.8792 on 3,080 Banking77 rows and says the Jev cell was not measured here

huggingface.co

GLiDE v1.2 prints Decision Index 0.4.1 Full 71.5, first of 229, at $0.04

huggingface.co

Span-02's card prints Decision Index 0.3 text index 63.62, and the blog prints thinking off at 61.2 against Jev 58.0 Blink v0.3 prints Decision Index 0.2.1 at 57.48

huggingface.co

Cloudflare posts Clef and Clef-flash, Apache 2.0 decision models on Workers AI llamacpp-jev answers /v1/systemone from an unmodified llama-server, including images

huggingface.co

hunch runs EmbeddingGemma2 as a CPU System One endpoint at 20 ms on text and 100 ms on images

huggingface.co

H2O-Lightning-4B v1.1 prints JevBench composite 72.5 against Jev at 71.5

huggingface.co

Base Labs posts mammoth-1 at Decision Index 0.4 Full 71.24; live 0.4.1 prints 68.4

huggingface.co

Base Labs posts mammoth-1 at Decision Index 0.4 Full 71.24; live 0.4.1 prints 68.4

huggingface.co

Strands Decider 2B scores 167 of 231 public JevBench tasks

huggingface.co

Open decision models start from an open language model Together prices Tev1 input at $0.04, and Ollama serves Choice, Score, and Noul

huggingface.co

imajev-4b: Hugging Face weights and JevBench 70.4

huggingface.co

Open decision models start from an open language model

huggingface.co

Open decision models start from an open language model

huggingface.co

JEV-27B-VL finishes 15 of 20 pick-and-place scenes

huggingface.co

Jared Palmer's Kev family copies Jev's API on Qwen, and trails it on new sources

huggingface.co

LlamaIndex's Logan Markewich ships jeff, a local System One API on GLiFormer

huggingface.co

Jeva-4B prints 0 flips on 600 Choice permutations, and BANKING77 at K=16 is 91.62 against Jev 82.50

huggingface.co

Open d1's d1-3B card prints 48.57 on a self-run of Decision Index 0.2.1 Liquid's chart puts d1 at 58.9 on an internal Decision Index 0.2.1

huggingface.co

Open d1's d1-3B card prints 48.57 on a self-run of Decision Index 0.2.1

huggingface.co

Open d1's d1-3B card prints 48.57 on a self-run of Decision Index 0.2.1

huggingface.co

meraGPT posts Decider 1 ahead of Jev on a teacher-labelled set

huggingface.co

Maincode's Matilda card prints 59.26 on a self-run of Decision Index 0.2.1

huggingface.co

Mizan 4B averages 78.7 on 3,000 Persian records

huggingface.co

PolicyLM-1.7B scores each policy category from 0 to 1

huggingface.co

Nace posts Drex 1.1, and the card titles the weights Drex DLM Nace posts Drex 1.5 at $0.04, and its chart shows 199 of 231 against Jev's 201

huggingface.co

Nace posts Drex 1.5 at $0.04, and its chart shows 199 of 231 against Jev's 201

huggingface.co

BranchOne returns Choice probabilities from Qwen3.5-0.8B-Base, median 0.305 s on ten short text questions

huggingface.co

Open-Jev 27B v1.1 answers 197 of 231 public JevBench tasks

huggingface.co

Open-Jev 27B v1.1 answers 197 of 231 public JevBench tasks

huggingface.co

Open-Jev 27B v1.1 answers 197 of 231 public JevBench tasks

huggingface.co

OpenBles Firewall 1 catches 119 of 120 agent attacks and flags no benign look-alike

huggingface.co

MiniCPM5-2B-Jev prints 78.79% on JevBench, first among open models at or under 2B

huggingface.co

hotdog-27B leads a 1,631-decision holdout and trails on 74 public items

huggingface.co

NIRNAY 450M prints 0.8792 on 3,080 Banking77 rows and says the Jev cell was not measured here

huggingface.co

NIRNAY 450M prints 0.8792 on 3,080 Banking77 rows and says the Jev cell was not measured here

huggingface.co

Blink v0.3 prints Decision Index 0.2.1 at 57.48 Decision Index 0.2 prints Jev at 51.67, and a later post names another first place

huggingface.co

Plumb-4B reaches 89 of 111 public hard items, and the official board scored it 65.84

huggingface.co

The Jeeves README prints 0.865 on the public JevBench hard tier

huggingface.co

Perplexity posted pplx-decider-v1.2 on October 16, with Decision Index 0.3 overall 68.09 on joint requests against v1.1 at 62.30

huggingface.co

Perplexity posted pplx-decider-v1.2 on October 16, with Decision Index 0.3 overall 68.09 on joint requests against v1.1 at 62.30

huggingface.co

Perplexity posted pplx-decider-v1.2 on October 16, with Decision Index 0.3 overall 68.09 on joint requests against v1.1 at 62.30 Decision Index 0.2 prints Jev at 51.67, and a later post names another first place

huggingface.co

Perplexity posted pplx-decider-v1.2 on October 16, with Decision Index 0.3 overall 68.09 on joint requests against v1.1 at 62.30

huggingface.co

PurpleMIST-Flash-1.0 scores 0.668 on 15,050 translated African-language decisions

huggingface.co

Opened October 19, JevBench ranks basal-omni-1.0-ultra first of 184 open systems on Capability at 82.5

huggingface.co

Span-02's card prints Decision Index 0.3 text index 63.62, and the blog prints thinking off at 61.2 against Jev 58.0

huggingface.co

WaterSheep fine-tunes ModernBERT and serves /v1/systemone with a multi-label question

huggingface.co

StartLux-Decision v1.1 27B Instant prints Full score 69.12, first of 174 on Decision Index 0.3.2

huggingface.co

StartLux-Decision v1.1 27B Instant prints Full score 69.12, first of 174 on Decision Index 0.3.2

huggingface.co

Sthānika's Sieve-31B card prints Decision Index 0.3.2 Full score 63.8

huggingface.co

ToddlerJev prints 68.70% on MMAU-mini, with weights under the Qwen Research License

huggingface.co

Vela 2.0's 9B card prints 41.63 on Decision Index 0.2.1

huggingface.co

Vela 2.0's 9B card prints 41.63 on Decision Index 0.2.1

huggingface.co

Vela 2.0's 9B card prints 41.63 on Decision Index 0.2.1

huggingface.co

Vela 2.0's 9B card prints 41.63 on Decision Index 0.2.1

huggingface.co

Vela 2.0's 9B card prints 41.63 on Decision Index 0.2.1

huggingface.co

Vela 2.0's 9B card prints 41.63 on Decision Index 0.2.1

huggingface.co

Opened October 26, Decision Index 0.4.1 Full score for d3 is 67.15; d3-turbo and d3-ultra have their own pages

huggingface.co

Opened October 26, Decision Index 0.4.1 Full score for d3 is 67.15; d3-turbo and d3-ultra have their own pages

huggingface.co

Opened October 26, Decision Index 0.4.1 Full score for d3 is 67.15; d3-turbo and d3-ultra have their own pages

huggingface.co

Opened October 26, Decision Index 0.4.1 Full score for d3 is 67.15; d3-turbo and d3-ultra have their own pages

huggingface.co

Opened October 26, Decision Index 0.4.1 Full score for d3 is 67.15; d3-turbo and d3-ultra have their own pages

huggingface.co

Opened October 26, Decision Index 0.4.1 Full score for d3 is 67.15; d3-turbo and d3-ultra have their own pages

huggingface.co

Opened October 26, Decision Index 0.4.1 Full score for d3-turbo is 67.31, and vision full is 72.46

huggingface.co

Opened October 26, Decision Index 0.4.1 Full score for d3-ultra is 70.18, second of 229

huggingface.co

A 400M encoder named Kev scores 79.96% against Jev on 17,476 of its own rows

huggingface.co

JevBench v1.5 ranks Cygnet 73.7 and Jev 1.13.0 72.1

huggingface.co

Mizan 4B averages 78.7 on 3,000 Persian records

huggingface.co

MiniCPM5-2B-Jev prints 78.79% on JevBench, first among open models at or under 2B

jacoblincool.github.io

StreamDecisionBench puts Kev-27B at 62.15 and Jev at 59.63

jamiewatters.work

Five runs leave a 3.3-point lead on 18 claim checks

jev-trader.vercel.app

A buy-or-sell every Monad block, with a mock default model

jevainews.com

Abe's 1,774 Snowflake posts cost $0.055 on Jev Noul against $1.87 on AI_FILTER

jevbench.app

jevbench.app replays recorded Jev answers on labeled examples, and one test is 20 of 21

jevbench.app

jevbench.app replays recorded Jev answers on labeled examples, and one test is 20 of 21

jevcli.vectorz.app

jevctl puts Jev on the command line, with a compaction hook that keeps lines verbatim

jevpilot.standardagents.ai

JevPilot is a playable driving sim that asks Jev which path to take

jevrail.samanp.xyz

Jevrail scores a shell command before it runs, and the printed corpus leaves the model off

jevsearch.fly.dev

jevsearch re-ranks keyword hits with Jev and scored 83% Hit@1 on the author's TypeSafe docs set

kev-eval.oaleinyk.xyz

CLM-8B caches the action list, and a fine-tuned head scores 31 of 38 DeepSWE holdouts Jared Palmer's Kev family copies Jev's API on Qwen, and trails it on new sources

labs.baseten.co

Base Labs posts mammoth-1 at Decision Index 0.4 Full 71.24; live 0.4.1 prints 68.4

langchain.com

LangChain ships a Jev classifier, then parks it on the agent loop

laya.convaiinnovations.com

Laya, an Apache 2.0 System One stack, publishes where it beats Jev's numbers and where it does not

laya.studio

Open decision models start from an open language model

laya.studio

Open decision models start from an open language model

learn.microsoft.com

Microsoft-Decision-1 lists $0.042 per million input tokens and calls itself 4.5 times quicker than Quyet-1.0-Large

levanto.ai

JevBench ranks Sage 1.3.0 first of 27 hosted systems at composite 74.0

levanto.ai

JevBench ranks Sage 1.3.0 first of 27 hosted systems at composite 74.0

liqiiiii.github.io

Jeva-4B prints 0 flips on 600 Choice permutations, and BANKING77 at K=16 is 91.62 against Jev 82.50

liquid.ai

Opened October 16, Artificial Analysis Decision Index prints d1-ultra at 65 and Jev 1.14 at 57

liquid.ai

Open decision models start from an open language model Open d1's d1-3B card prints 48.57 on a self-run of Decision Index 0.2.1

lkranz.com

Jevflake puts Jev in Snowflake: noul, choice, and score as SQL functions

logicrw.github.io

A week of Jev repos: routers, MCP, graphs, games

lyzr.ai

CLERC top-1 moves from 5% to 18% when jev-1.12 reranks the shortlist

meragpt.com

meraGPT posts Decider 1 ahead of Jev on a teacher-labelled set

motherduck.com

MotherDuck's prompt_jev() labels 100,000 AG News rows in 40 seconds for $0.50

n.demir.io

BranchOne returns Choice probabilities from Qwen3.5-0.8B-Base, median 0.305 s on ten short text questions

nace.ai

Nace posts Drex Max at Decision Index 0.3.2 public 67.59, $0.06 per million input Nace posts Drex 1.5 at $0.04, and its chart shows 199 of 231 against Jev's 201 Decision Index 0.2 prints Jev at 51.67, and a later post names another first place

nace.ai

Decision Index 0.2 prints Jev at 51.67, and a later post names another first place

netlify.com

OpenRouter, Cloudflare, and Netlify start serving Jev

npmjs.com

toolgate 0.15 separates writing a script from running it, 19 of 20 on the paired set

npmjs.com

On 40 client requests, Jev and Opus both scored 37 of 40, at $0.0015 against $0.005 to $0.015

npmjs.com

An n8n community node routes Jev answers onto category branches, including Needs Review

ollama.com

MiniCPM5-2B-Jev prints 78.79% on JevBench, first among open models at or under 2B

ollama.com

Bespoke Nimble 9B matches 90% of its own labels and 74.8% of a public suite

ollama.com

Bespoke Nimble 9B matches 90% of its own labels and 74.8% of a public suite

ollama.com

Together prices Tev1 input at $0.04, and Ollama serves Choice, Score, and Noul

openai.com

OpenAI Decisions API is generally available; this desk's October 23 POSTs returned HTTP 200 OpenAI's Decisions API preview says results come back in 150 milliseconds

openai.com

OpenAI Decisions API is generally available; this desk's October 23 POSTs returned HTTP 200

openrouter.ai

On 492 tool calls, OpenRouter Armor's default Jev 1.13 scored 90.7%, 446 of 492

openrouter.ai

On October 26 TypeSafe's jev-latest named jev-1.14.0, and both aliases point at 1.14.0 OpenRouter, Cloudflare, and Netlify start serving Jev

openrouter.ai

Arena scores typesafe/jev-router at +3.6% and $0.155 a task

openrouter.ai

Opened October 16, Artificial Analysis Decision Index prints d1-ultra at 65 and Jev 1.14 at 57

openrouter.ai

Perplexity posted pplx-decider-v1.2 on October 16, with Decision Index 0.3 overall 68.09 on joint requests against v1.1 at 62.30

openrouter.ai

Perplexity posted pplx-decider-v1.2 on October 16, with Decision Index 0.3 overall 68.09 on joint requests against v1.1 at 62.30

openrouter.ai

On October 26 TypeSafe's jev-latest named jev-1.14.0, and both aliases point at 1.14.0

openrouter.ai

On October 26 TypeSafe's jev-latest named jev-1.14.0, and both aliases point at 1.14.0

openrouter.ai

Arena scores typesafe/jev-router at +3.6% and $0.155 a task

openrouter.ai

Arena scores typesafe/jev-router at +3.6% and $0.155 a task OpenRouter, Cloudflare, and Netlify start serving Jev

openrouter.ai

Arena scores typesafe/jev-router at +3.6% and $0.155 a task

openrouter.ai

Inception calls Mercury Decide first on OpenRouter's JevBench v1.4, and prints no score

openrouter.ai

Inception calls Mercury Decide first on OpenRouter's JevBench v1.4, and prints no score

openrouter.ai

Microsoft-Decision-1 lists $0.042 per million input tokens and calls itself 4.5 times quicker than Quyet-1.0-Large

openrouter.ai

Span-02's card prints Decision Index 0.3 text index 63.62, and the blog prints thinking off at 61.2 against Jev 58.0

pc.watch.impress.co.jp

On October 26 TypeSafe's jev-latest named jev-1.14.0, and both aliases point at 1.14.0

pgjev.com

Zachi's pg-jev judges 2,000 Postgres rows in about 3.5 seconds

pgjev.com

Zachi's pg-jev judges 2,000 Postgres rows in about 3.5 seconds

pgjev.zachi.dev

Zachi's pg-jev judges 2,000 Postgres rows in about 3.5 seconds

poker.ccswitch.online

A Hold'em table asks Jev about the opponent and lets code pick the bet

polylane.com

Polylane moved five on-call decisions to Jev and prints a 39.1% lower bill per 1,000 calls

portal.sina.com.hk

Sthānika's Sieve-31B card prints Decision Index 0.3.2 Full score 63.8

position.secondfirst.ai

Jevsus ran a true/false diode on jev-1.13.0, and 45% of statements move with the wording

pub.dev

Jevis is a Flutter integration-test agent that only taps what you registered

pypi.org

On DeepEval's Paris weather example, jev-1.13.0 scored 0.254 when the answer invented a breeze

pypi.org

TypeLLM scores Qwen3.8-27B at 228 of 231 public JevBench tasks when thinking is on

qdrant.tech

On WANDS, more Jev questions move the mean to 0.610 and leave the median at 0.561

qiita.com

Abe's 1,774 Snowflake posts cost $0.055 on Jev Noul against $1.87 on AI_FILTER

qiita.com

On October 26 TypeSafe's jev-latest named jev-1.14.0, and both aliases point at 1.14.0

quirkyagents.com

TypeLLM said Jev 1.14 cannot return every vehicle box, including Tesla's flying Roadster; this desk's every-vehicle Box was one rectangle, n = 3, and type array returned HTTP 400

quyet.ai

Quyet-1.0-Large's homepage prints JevBench v1.6.0 at 71.39

respan.ai

Span-02's card prints Decision Index 0.3 text index 63.62, and the blog prints thinking off at 61.2 against Jev 58.0

respan.ai

Span-02's card prints Decision Index 0.3 text index 63.62, and the blog prints thinking off at 61.2 against Jev 58.0

search.maven.org

Spring publishes a Java client that asks Jev before the chat model runs

securitylabs.datadoghq.com

Datadog prints Jev Noul recall 82.90% to 3.88% under a counterfeit policy, and this desk's is_risky fell from 0.94 to 0.13

semgate-example-90572511273.asia-northeast1.run.app

semgate asks Jev about each HTTP request, then a live range to attack it

share.querystory.ai

TechCrunch prices one Jev monitor at $2.94, and the demo page prints another estimate

siliconangle.com

TechCrunch covers Jev, and quotes a Vercel engineer who swapped Luna for it

simple-jev.featherless.ai

SimpleJev turns an open model into a classifier by reading the next-token logits

softwaredoug.com

On WANDS, more Jev questions move the mean to 0.610 and leave the median at 0.561

sosyaldijital.com

On October 26 TypeSafe's jev-latest named jev-1.14.0, and both aliases point at 1.14.0

spring-ai-community.github.io

Spring publishes a Java client that asks Jev before the chat model runs

spring-ai-community.github.io

Spring publishes a Java client that asks Jev before the chat model runs

spring.io

Spring publishes a Java client that asks Jev before the chat model runs

sqwish.ai

Opened October 17, JevBench API board v1.7.46 ranks kettei-fat first of 39 at composite 77.8

sregym.com

SREGym put Jev on a Codex agent: 24 passes out of 50, up from 20

sthanika.ai

Sthānika's Sieve-31B card prints Decision Index 0.3.2 Full score 63.8

strandsagents.com

Strands Decider 2B scores 167 of 231 public JevBench tasks

subq.ai

SubQ Snap 1.1 prints 65 on Artificial Analysis Decision Index, $0.04 per million input

subq.ai

SubQ Snap 1.1 prints 65 on Artificial Analysis Decision Index, $0.04 per million input

sumleo.github.io

RLCDAlignBench puts a generic Noul at median AUROC 0.886, and median ECE at 0.168

system-one-security-triage.rzuniga-9b4.workers.dev

A 100-case security triage demo put Jev at 65.3% balanced accuracy against Terra 75.1% and Opus 91.2%

techcrunch.com

TechCrunch covers Jev, and quotes a Vercel engineer who swapped Luna for it

techcrunch.com

Vercel's October 2026 AI Gateway Production Index prints Jev at 35.3% of paid teams in its first two weeks

techcrunch.com

TechCrunch covers Jev, and quotes a Vercel engineer who swapped Luna for it

techcrunch.com

TechCrunch prices one Jev monitor at $2.94, and the demo page prints another estimate OpenAI's Decisions API preview says results come back in 150 milliseconds

teenage.works

ToddlerJev prints 68.70% on MMAU-mini, with weights under the Qwen Research License

the-decoder.com

OpenAI Decisions API is generally available; this desk's October 23 POSTs returned HTTP 200

theinformation.com

TechCrunch covers Jev, and quotes a Vercel engineer who swapped Luna for it

thenewstack.io

OpenAI's Decisions API preview says results come back in 150 milliseconds

theregister.com

HiddenLayer prints Jev at 91% Confirm Price on a checkout ad, and this desk's jev-1.13.0 still chose place_order at 0.89

thoughtworks.com

CLERC top-1 moves from 5% to 18% when jev-1.12 reranks the shortlist

together.ai

Together prices Tev1 input at $0.04, and Ollama serves Choice, Score, and Noul

together.ai

Open decision models start from an open language model

trueforge.dev

On 40 client requests, Jev and Opus both scored 37 of 40, at $0.0015 against $0.005 to $0.015

trueforge.dev

On 40 client requests, Jev and Opus both scored 37 of 40, at $0.0015 against $0.005 to $0.015

typellm.ai

TypeLLM scores Qwen3.8-27B at 228 of 231 public JevBench tasks when thinking is on

typellm.ai

TypeLLM scores Qwen3.8-27B at 228 of 231 public JevBench tasks when thinking is on

typellm.ai

TypeLLM scores Qwen3.8-27B at 228 of 231 public JevBench tasks when thinking is on

typellm.ai

TypeLLM scores Qwen3.8-27B at 228 of 231 public JevBench tasks when thinking is on

typellm.ai

TypeLLM scores Qwen3.8-27B at 228 of 231 public JevBench tasks when thinking is on

typellm.ai

TypeLLM scores Qwen3.8-27B at 228 of 231 public JevBench tasks when thinking is on

typellm.ai

TypeLLM scores Qwen3.8-27B at 228 of 231 public JevBench tasks when thinking is on

typellm.ai

TypeLLM said Jev 1.14 cannot return every vehicle box, including Tesla's flying Roadster; this desk's every-vehicle Box was one rectangle, n = 3, and type array returned HTTP 400 TypeLLM scores Qwen3.8-27B at 228 of 231 public JevBench tasks when thinking is on

typellm.ai

TypeLLM said Jev 1.14 cannot return every vehicle box, including Tesla's flying Roadster; this desk's every-vehicle Box was one rectangle, n = 3, and type array returned HTTP 400 TypeLLM scores Qwen3.8-27B at 228 of 231 public JevBench tasks when thinking is on

typellm.ai

TypeLLM said Jev 1.14 cannot return every vehicle box, including Tesla's flying Roadster; this desk's every-vehicle Box was one rectangle, n = 3, and type array returned HTTP 400 TypeLLM scores Qwen3.8-27B at 228 of 231 public JevBench tasks when thinking is on

unsloth.ai

Open decision models start from an open language model

vals.ai

Opened October 16, Artificial Analysis Decision Index prints d1-ultra at 65 and Jev 1.14 at 57

venice.ai

Venice lists Jev on its API: same key, no copy of the state

vercel.com

Vercel's form router keeps Jev's destination when confidence is at least 0.95

vercel.com

Vercel's October 2026 AI Gateway Production Index prints Jev at 35.3% of paid teams in its first two weeks Stagehand, Chrome, and a11y trees: computer use as a Jev question TypeSafe starts pulling people off the Jev waitlist

vercel.com

Vercel's October 2026 AI Gateway Production Index prints Jev at 35.3% of paid teams in its first two weeks

vercel.com

Vercel's October 2026 AI Gateway Production Index prints Jev at 35.3% of paid teams in its first two weeks

vercel.com

Opened October 16, Artificial Analysis Decision Index prints d1-ultra at 65 and Jev 1.14 at 57

vercel.com

Base Labs posts mammoth-1 at Decision Index 0.4 Full 71.24; live 0.4.1 prints 68.4

vercel.com

Opened October 16, Artificial Analysis Decision Index prints d1-ultra at 65 and Jev 1.14 at 57

vercel.com

Vercel's October 2026 AI Gateway Production Index prints Jev at 35.3% of paid teams in its first two weeks

violoop.ai

Violoop ran the same desktop harness with Jev and with Opus 5.5 on one QuickBooks bill

vllm-sr.ai

Vela 2.0's 9B card prints 41.63 on Decision Index 0.2.1

xn--grn-sna.name

Arena scores typesafe/jev-router at +3.6% and $0.155 a task Jared Palmer's Kev family copies Jev's API on Qwen, and trails it on new sources OpenRouter, Cloudflare, and Netlify start serving Jev

youtube.com

One builder ran Jev on 1,500 of his own emails

zefan-cai.github.io

Open-Jev 27B v1.1 answers 197 of 231 public JevBench tasks

zefan-cai.github.io

Open-Jev 27B v1.1 answers 197 of 231 public JevBench tasks