Jev 1.14

+ contribute

Jev 1.14 is a closed-weights decision model from TypeSafe, released 14 October 2026. Licence: proprietary.

Catalogue identity notes

Exact Decision Index tested engine jev; reported model Jev 1.14. Type: None; technique: None. Comparison entry; no current run command, API price or speed inferred from the benchmark. Jev 1.14 is available but remains unscored: the licensed per-benchmark export omits its rows. Jev 1.13 measurements are not substituted.

Model information sources

Release date ↗

64.2±1.3#63 of 355 decision modelsfrom 37 benchmark results across 7 sources

$ how it compares

The decision models ranked either side of Jev 1.14, out of 357 with a score, then more from TypeSafe.

#	model	score	where	released
61	d1 (hosted) Liquid AI	65±1	api price unknown
62	KnowLine 4B Gen5.5 PelaAI	64±1	model files	2026-10-22
63	Jev 1.14 TypeSafe	64±1	api $0.042/M input tokens	2026-10-14
64	Jeffan-4B-2010 tattrongvu	64±1	model files
65	bosun-v3.1-0.6b (DecisionBench aaa9dd06) Hanno Labs	64±2	model files	2026-09-22
	more from TypeSafe
49	Jev 1.13 TypeSafe	66±1	api $0.042/M input tokens	2026-09-15

see all decision models · find the ones that fit your machine

$ what it's good at

retrieval & classification▁▁▁·▁▁▁▁▃▃█▅ #106 of 258 · 10% of its score · 6 tests

Beats AutoTrust-JEV-27B v1.1, Decider 31B (NVFP4), Quyet-1.0-Large here, though they rank higher overall. Trails vLLM-SR vela3-fast (Vela 3.0, 0.4B), crystal-0.8b, Jev-Style 2B v4 here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
System One Bench: MIRACL pairwise (Arabic)	90.0% ✎	▂···▁··▃▄▂█▇ beats 21/29	easy · avg 84%	▇ 3.4%	typical
System One Bench: MASSIVE intent (Arabic)	85.0% ✎	▄·▁·▁▃▃▃█▄▇▇ beats 16/29	easy · avg 76%	▇ 3.0%	typical
System One Bench: MIRACL rerank (Arabic)	87.5% ✎	▁▂·▁▁·▂·▄▄█▇ beats 16/29	easy · avg 79%	▇ 1.9%	typical
System One Bench: MASSIVE scenario (Arabic)	65.0% ✎	▃▁·▄▃▃█▂▁▁·▃ beats 12/29	medium · avg 60%	▇ 1.3%	at its level
System One Bench: AG News (English)	89.0% ✎	▁···▅▁▅▂▆█▂▅ beats 16/29	easy · avg 88%	▇ 0.4%	blunt
System One Bench: Banking77 (English)	78.5% ✎	█▅▆█▅▃▃▃···▂ beats 1/26	very easy · avg 87%	▇ 0.0%	blunt · noisy

language understanding▂▂▂▃▂▅▆▃█▆▄▃ #80 of 256 · 9.2% of its score · 9 tests

Beats Turion Jev 31B, AJev lm6q (Gemma 4 31B), oenogarum (Gemma 4 31B) here, though they rank higher overall. Trails Sieve 4B, Amber 4B, Decider 35B-A3B here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
System One Bench: BoolQ (English)	90.0% ✎	▁▁▁▃·▁▃▃▇▇█▅ beats 18/29	easy · avg 86%	▇ 2.3%	typical
System One Bench: OSACT4 offensive (Arabic)	87.0% ✎	▁······▁▃█▇▄ beats 10/29	easy · avg 84%	▇ 2.1%	sharp
System One Bench: SST-2 (English)	95.0% ✎	▁·▃▁▂▂▃▅█·▂▁ beats 17/29	easy · avg 92%	▇ 1.4%	typical
System One Bench: SST-5 (English)	60.0% ✎	▃·▁▁▃·▇▄▄▄█▅ beats 25/29	medium · avg 51%	▇ 1.4%	blunt · at its level
System One Bench: XNLI Arabic (Arabic)	71.3% ✎	▄▁▃▁▄▇█▅▃▁·▃ beats 11/29	medium · avg 68%	▇ 0.7%	at its level
System One Bench: XNLI French (French)	76.7% ✎	▂▅·▁▁▅▅█▃▁·▃ beats 8/29	medium · avg 76%	▇ 0.6%	blunt · at its level
System One Bench: Tweet sentiment (Arabic)	76.0% ✎	▁······▁▁▄█▄ beats 27/29	medium · avg 62%	▇ 0.5%	blunt · at its level
System One Bench: XNLI English (English)	86.0% ✎	▃·▁·▃▇█▄█▃▃▁ beats 11/29	easy · avg 85%	▇ 0.2%	blunt · noisy

1 more test

test	result	vs the others in its run	difficulty	counts	why
System One Bench: DAIR Emotion (English)	57.5% ✎	██▃·····▁··▁ beats 17/29	medium · avg 59%	▇ 0.0%	blunt · at its level

knowledge & reasoning▁▂▂▂▂▃▃▅▅▆█▂ #55 of 263 · 25% of its score · 6 tests

Beats OneJev1.6-27B, baselabs elephant-1 (9B), StartLux-Decision v1.1 9B here, though they rank higher overall. Trails Solar Decide, Clef-flash, Clef-omni here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
Etsabary · 1342 text decisions	76.8% ✎	▁··▂·▃▅▁▃▃▂█ beats 14/24	medium · avg 47%	▇▇▇▇▇▇ 16%	sharp · at its level
CaseHOLD · missing authority	63.9% ✎	▄···▄······█ beats 1/3	medium · avg 40%	▇▇ 5.5%	sharp · at its level
CONFLICTS · source conflict types	59.6% ✎	▃▃··▅▃···▃▃█ beats 1/9	medium · avg 45%	▇▇ 4.2%	at its level
Clinical criteria · TrialGPT	65.2% ✎	no other model yet		—	not counted yet: no other model in the same run to compare it with
Clinical criteria · MIMIC-III	76.8% ✎	no other model yet		—	not counted yet: no other model in the same run to compare it with
Clinical criteria · MIMIC-IV	80.7% ✎	no other model yet		—	not counted yet: no other model in the same run to compare it with

vision▁·▃▂▃▃▇▄▄█▄▃ #5 of 79 · 31% of its score · 10 tests

Beats Fastino GLiDE 1.2, StartLux-Decision v1.1 27B, vLLM-SR d3-ultra (Decision 3.0, 27B) here, though they rank higher overall.

test	result	vs the others in its run	difficulty	counts	why
Etsabary · 200 image decisions	70.5% ✎	▂▂··▂▃▂·█▃▂▂ beats 6/14	medium · avg 45%	▇▇▇▇ 11%	sharp · at its level
System One Bench: Image object count	93.5% ✎	▃····▃█▃▅·█▃ beats 9/11	easy · avg 71%	▇▇ 4.5%	sharp
System One Bench: SST-2 screenshot	93.5% ✎	█···███·█·▄▄ beats 5/11	easy · avg 87%	▇ 3.7%	sharp
System One Bench: Visual yes/no questions	86.5% ✎	▂······▃█▃·▃ beats 2/11	medium · avg 80%	▇ 3.3%	at its level
System One Bench: Arabic sentiment screenshot	63.5% ✎	▄··▄··▂···▆█ beats 9/11	very hard · avg 37%	▇ 2.9%	sharp · at its level · noisy
System One Bench: Oxford pets classification	88.5% ✎	███·█▄▄·▄··▄ beats 9/11	easy · avg 77%	▇ 2.8%	typical
JevalKit: VisualDecisionBench images	85.1% ✎	▄▄···█▄·█▄█▄ beats 9/10	easy · avg 78%	▇ 1.3%	blunt
Benchmark Heaven imagejevbench v0.3.0: intelligence (points)	49.8907 points ✎	▃·▃▃··▅█▅▅·▃ beats 11/12	avg 32.235 points	▇ 0.9%	blunt

2 more tests

test	result	vs the others in its run	difficulty	counts	why
Benchmark Heaven imagejevbench v0.3.0: calibration (points)	70.5406 points ✎	▃···▃·▅█▅▃▃▅ beats 4/12	avg 68.584 points	▇ 0.4%	blunt
System One Bench: RefCOCO grounding (IoU ≥ 0.5)	78.5% ✎	no other model yet		—	not counted yet: no other model in the same run to compare it with

workflows▁▁▁▂▁▂▂▆█▄▁▁ #15 of 114 · 12% of its score · 5 tests

Beats Nace.AI Drex Max v1 (27B), d1-ultra, baselabs elephant-1 (9B) here, though they rank higher overall. Trails GPT-6 Luna Decisions, reflex-27b (DecisionBench 1d4bf0f2), Microsoft-Decision-1 here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
JevalKit: Procedural Typed Decisions	82.9% ✎	█·███▄·▄▄▄·▄ beats 9/12	easy · avg 67%	▇▇ 4.7%	sharp
System One Bench: Typed decisions (English)	74.2% ✎	▁▁·▁▁▁▅▄▇██▁ beats 23/29	medium · avg 63%	▇ 3.1%	blunt · at its level
JevalKit: DecisionBench standard eval	73.7% ✎	▃▃·▃···▅█▅·█ beats 4/12	easy · avg 70%	▇ 2.3%	typical
JevalKit: JevBench public 231	87.4% ✎	█▄▄▄▄▄██▄··▄ beats 8/12	easy · avg 82%	▇ 1.2%	typical
JevalKit: Typed Decisions held-out modal labels	74.6% ✎	▃▅··▃·▃▃·█▃█ beats 4/12	easy · avg 74%	▇ 0.1%	blunt

other tests5 not in a task · 13% of its score

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: private same skills (equated points)	58.9100 points ✎	▂▂▂▃▃▃▄▅▅▇█▃ beats 203/247	avg 41.206 points	▇▇ 5.5%	blunt
Decision Index 0.4.1: private new domains (equated points)	62.2400 points ✎	▃█▃▃▂▆▃▃▄▄▄▄ beats 193/247	avg 38.861 points	▇ 3.0%	blunt
JevalKit: JevAlt-Bench EN/TR/DE	90.5% ✎	▅··▃▃▅▃█··▅▃ beats 9/12	very easy · avg 87%	▇ 2.7%	maxed out
JevalKit: Decision Index public 0.3 (official weighted points)	60.8000 points ✎	▃▃·▃·▃▃▃▅▅·█ beats 9/12	avg 39.649 points	▇ 1.3%	blunt
JevalKit: Chinese decision validation v0.2	88.4% ✎	▃····█·█▃▃█▃ beats 4/12	very easy · avg 88%	▇ 0.2%	blunt

$ where to run it

Through an API

api · $0.042/M input tokens · docs.typesafe.ai ↗ · price sourceObserved 2026-10-31: $0.042 per million input tokens; output tokens free. Versioned model ID jev-1.14.0.

Model repository and documentation: docs.typesafe.ai ↗

$ how the score was worked out · System One Bench (Ali Alhashi… 40%, Etsabary · decision benchmarks 26%, JevalKit (Valen Team) 14% and 4 more

Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works

source	results	share of the score	licence
System One Bench (Ali Alhashimi)	21	▇▇▇▇▇▇▇▇▇▇▇▇ 40%	CC-BY-4.0
Etsabary · decision benchmarks	2	▇▇▇▇▇▇▇▇ 26%	MIT
JevalKit (Valen Team)	8	▇▇▇▇ 14%	Apache-2.0
Decision Index 0.4.1	2	▇▇▇ 9%	CC BY 4.0
Commit or Abstain? System One Decision Models, LLMs, and Missing Legal Authority	1	▇▇ 6%	CC-BY-4.0
Do We Need an LLM to Know Why Sources Conflict? Decision Models for Knowledge-Conflict Typing in RAG	1	▇ 4%	CC-BY-4.0
Benchmark Heaven (own decision evaluations)	2	▇ 1%	CC BY 4.0
prior · pull toward the middle		▇ 1%
weighted average		= 64.2

not counted (4)

System One Bench (Ali Alhashimi)	System One Bench: RefCOCO grounding (IoU ≥ 0.5) [correct]	78.5%	fewer than three models have this test yet, or this source ran it for this model alone
Can JEV Support Clinical Trial Criteria Assessment?	Clinical criteria · TrialGPT [correct]	65.2%	fewer than three models have this test yet, or this source ran it for this model alone
Can JEV Support Clinical Trial Criteria Assessment?	Clinical criteria · MIMIC-III [correct]	76.8%	fewer than three models have this test yet, or this source ran it for this model alone
Can JEV Support Clinical Trial Criteria Assessment?	Clinical criteria · MIMIC-IV [correct]	80.7%	fewer than three models have this test yet, or this source ran it for this model alone

$ improve this page

Update model information Add a place to run it Add a test result Propose a benchmark Correct a benchmark score Edit data on GitHub

Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.

data updated 2026-11-03 02:37 UTC · sources and licences · json