Kev 4B r10

+ contribute

Kev 4B r10 is an open-weights decision model from Jared Palmer, released 24 September 2026. Its published size is 4.66B parameters. Licence: apache-2.0.

Catalogue identity notes

Exact Decision Index tested engine kev-4b-r10; reported model Kev 4B r10. Type: LoRA + head; technique: autoregressive. No API price or speed inferred from the benchmark. Current repository metadata checked separately; an unpinned alias is not evidence that today’s weights reproduce this historical run. Reviewed Ollaya local route added 2026-10-31: kev:4b, manifest c0e732655dc9d799dd21ba7f2f9a445fbd50ee83578e96ccfbf691c1be9dde53, matching the independent accuracy-run manifest. Native family implementation; published runtime precision label F32. Memory is measured peak GPU increase (11200 MiB, rounded upward to 11.0 GiB) on RTX 5090 during the short five-question triage workload, not download size or a maximum-context guarantee. Runtime installation is required; no fresh execution performed here. Historical benchmark adapter and input protocol still vary by result.

Model information sources

Release date ↗ · Weights licence ↗ · Parameter count ↗

Publisher parameter count; includes the served backbone, not just adapter size.

48.5±0.9#217 of 355 decision modelsfrom 79 benchmark results across 4 sources

$ how it compares

The decision models ranked either side of Kev 4B r10, out of 357 with a score, then more from Jared Palmer.

#	model	score	where	RAM	released
217	Aurelis-2B Leptos Labs	49±1	model files
218	zerank-2-reranker (DecisionBench 5eae30d5) zeroentropy	49±2	model files
219	Kev 4B r10 Jared Palmer	49±1	local	~11GB	2026-09-24
220	StartLux-Decision v1.1 0.8B StartLux	49±1	model files
221	mmastrac diffgemma (decision readout) vllm-project	48±1	source code
	more from Jared Palmer
88	Kev 27B Jared Palmer	62±1	model files		2026-09
159	kev-4b (DecisionBench 485ace87) Jared Palmer	55±2	model files
162	kev-8b (DecisionBench c80773da) Jared Palmer	55±2	model files

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

$ what it's good at

tools & automation▁▃▁▁▁▂▃▄▆██▆ #141 of 250 · 5.1% of its score · 6 tests

Beats Cerno 4B v1.1, Trio-Spark DI (9B), Decider 35B-A3B here, though they rank higher overall. Trails JevK5 Lite, Amber 0.4B, Crivo-1B here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: BFCL (case exact accuracy) submitted	93.1% ✎	▁▂▁▁▁▂▂▄▅▆█▁ beats 110/246	very easy · avg 92%	▇ 1.7%	sharp · noisy
Decision Index 0.4.1: API-Bank (accuracy) submitted	53.3% ✎	▁▁▁▁▁▁▁▂▃▂█▂ beats 58/236	easy · avg 70%	▇ 1.5%	sharp · at its level · noisy
Decision Index 0.4.1: When2Call (accuracy) submitted	66.7% ✎	▁▁▁▂▃▅▅▆█▇█▃ beats 84/246	easy · avg 72%	▇ 0.7%	at its level · noisy
Decision Index 0.4.1: Home appliance simulator (case exact accuracy) submitted	14.8% ✎	█▁▁▃▄▄▆▄▁▂▁▃ beats 79/243	medium · avg 37%	▇ 0.6%	sharp · at its level · noisy
Decision Index 0.4.1: ToolRet (nDCG@10) submitted	61.0% ✎	▁▁▁▁▁·▁▁▁▄█▁ beats 74/246	medium · avg 60%	▇ 0.6%	blunt · at its level
Decision Index 0.4.1: RouterBench (selected quality (quality objective)) submitted	80.0% ✎	no other model yet		—	failed run: Publisher excludes this protocol because its prompt reveals GPT-4 as the best route. Evidence: https://huggingface.co/spaces/multimodalart/jev-decision-index/blob/c130faf65f82e81d3bdefb01e758c712dc5c5199/data/index.json

retrieval & classification▁▁▁·▁▁▁▁▃▃█▅ #192 of 258 · 11% of its score · 13 tests

Beats JevK5 9B v0.3.3, Hopper (G) 1.2 Qwen3.5-4B LoRA here, though they rank higher overall. Trails Bosun v3.1 0.6B LoRA, Mola-68M, Bosun v3.1 1.7B LoRA here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
System One Bench: MASSIVE intent (Arabic)	76.5% ✎	▄·▁·▁▃▃▃█▄▇▇ beats 6/29	easy · avg 76%	▇ 2.5%	at its level
System One Bench: MIRACL pairwise (Arabic)	86.5% ✎	▂···▁··▃▄▂█▇ beats 9/29	easy · avg 84%	▇ 2.5%	at its level
System One Bench: MIRACL rerank (Arabic)	79.5% ✎	▁▂·▁▁·▃·▄▄█▇ beats 6/29	easy · avg 79%	▇ 1.5%	at its level
Decision Index 0.4.1: BRIGHT (nDCG@10) submitted	39.0% ✎	▁▁▁▁▁▁▁▂▂▄▇█ beats 93/246	medium · avg 38%	▇ 1.3%	blunt · at its level
Decision Index 0.4.1: Amazon ESCI (macro-F1) submitted	39.7% ✎	▁▁▁▁▁▁▂▂▃▃█▅ beats 59/246	medium · avg 47%	▇ 1.2%	at its level
System One Bench: MASSIVE scenario (Arabic)	57.0% ✎	▃▁·▄▃▂█▂▁▁·▃ beats 6/29	medium · avg 60%	▇ 0.8%	at its level
Decision Index 0.4.1: HoVer (accuracy) submitted	59.7% ✎	▆▆▇▅▄▄▃▇▄█▆▄ beats 47/246	easy · avg 74%	▇ 0.4%	blunt · at its level
Decision Index 0.4.1: PhishNChips phishing decisions (accuracy) submitted	55.3% ✎	█▅▅▄▃▂▁▁▁▁·▁ beats 71/245	easy · avg 70%	▇ 0.3%	at its level · noisy

5 more tests

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: CLINC150+OOS (macro-F1) submitted	75.0% ✎	▁▁▁▁▁▁▁▂▃▆█▄ beats 50/238	easy · avg 83%	▇ 0.3%	sharp · noisy
System One Bench: AG News (English)	90.5% ✎	▁···▅▁▅▂▆█▂▅ beats 21/29	easy · avg 88%	▇ 0.2%	blunt
Decision Index 0.4.1: BANKING77 (macro-F1) submitted	82.9% ✎	▁·▁▁▁▁▁▁▂█▇█ beats 119/238	easy · avg 79%	▇ 0.2%	noisy
System One Bench: Banking77 (English)	88.0% ✎	█▅▆█▅▃▃▃···▂ beats 13/26	very easy · avg 87%	▇ 0.0%	blunt · noisy
Decision Index 0.4.1: SGD/SGD-X (macro-F1) submitted	47.8% ✎	no other model yet		—	failed run: Publisher reports a builder bug exposing dialogue turns after the classified turn; awaiting corrected rerun. Evidence: https://huggingface.co/spaces/multimodalart/jev-decision-index/blob/c130faf65f82e81d3bdefb01e758c712dc5c5199/data/index.json

language understanding▂▂▂▃▂▅▆▃█▆▄▃ #175 of 256 · 21% of its score · 18 tests

Beats Nari Decision 4B v3.2 here, though they rank higher overall. Trails DecisionTune 1.0, jdi-ptr14-0.8b, FLock this-that 1.2 here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: HellaSwag (accuracy) submitted	78.9% ✎	▂▄▂▂▄▇▃▇█▅▃▁ beats 80/246	very easy · avg 85%	▇▇ 4.3%	sharp
Decision Index 0.4.1: NLI4CT (macro-F1) submitted	74.6% ✎	▁▁▁▂▂▃▃▄▇▇█▄ beats 97/246	easy · avg 74%	▇ 3.4%	at its level
Decision Index 0.4.1: ANLI (macro-F1) submitted	48.1% ✎	▁▂▃▅▅█▂····▁ beats 57/246	medium · avg 60%	▇ 3.1%	at its level
System One Bench: OSACT4 offensive (Arabic)	88.5% ✎	▁······▁▃█▇▄ beats 12/29	easy · avg 84%	▇ 1.8%	sharp · at its level
System One Bench: BoolQ (English)	91.5% ✎	▁▁▁▃·▁▃▃▇▇█▅ beats 24/29	easy · avg 86%	▇ 1.7%	at its level
Decision Index 0.4.1: iSarcasmEval (Sarcasm F1 · track A, English) submitted	46.8% ✎	▁·▁·▁▁▂▄▅█▇▄ beats 118/246	medium · avg 44%	▇ 1.2%	at its level
System One Bench: SST-2 (English)	92.5% ✎	▁·▃▁▂▃▃▅█·▂▁ beats 8/29	easy · avg 92%	▇ 1.2%	typical
System One Bench: SST-5 (English)	52.0% ✎	▃·▁▁▃·▇▄▄▄█▅ beats 10/29	medium · avg 51%	▇ 0.7%	blunt · at its level

10 more tests

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: ContractNLI (macro-F1) submitted	64.4% ✎	▁▁▁▁▁▁▁▂▃▇▇█ beats 56/246	easy · avg 71%	▇ 0.7%	at its level · noisy
Decision Index 0.4.1: VAST (macro-F1) submitted	52.4% ✎	▁▁▁▂▅▄▄▂▄▆█▄ beats 85/246	medium · avg 62%	▇ 0.6%	at its level
Decision Index 0.4.1: ACOS (per-review F1) submitted	9.8% ✎	▁··▂▃▃▅▇█▇▃▃ beats 52/246	hard · avg 22%	▇ 0.5%	noisy
Decision Index 0.4.1: RAGTruth (F1 on hallucinated class) submitted	67.5% ✎	▁▁▁▁▁▁▁▁▁▃▆█ beats 86/246	easy · avg 66%	▇ 0.5%	at its level · noisy
Decision Index 0.4.1: FinEntity (macro-F1) submitted	87.1% ✎	▁···▁▁▁▁▃▆█▃ beats 113/246	easy · avg 84%	▇ 0.5%	noisy
System One Bench: XNLI Arabic (Arabic)	72.7% ✎	▄▁▃▁▄▇█▅▃▁·▃ beats 14/29	medium · avg 68%	▇ 0.4%	at its level
System One Bench: XNLI French (French)	79.3% ✎	▂▅·▁▁▅▅█▃▁·▃ beats 12/29	medium · avg 76%	▇ 0.4%	blunt · at its level
System One Bench: Tweet sentiment (Arabic)	63.0% ✎	▁······▁▁▄█▄ beats 7/29	medium · avg 62%	▇ 0.3%	blunt · at its level
System One Bench: XNLI English (English)	90.7% ✎	▃·▁·▃▇█▄█▃▃▁ beats 21/29	easy · avg 85%	▇ 0.1%	blunt · noisy
System One Bench: DAIR Emotion (English)	54.5% ✎	██▃·····▁··▁ beats 4/29	medium · avg 59%	▇ 0.0%	blunt · at its level

arts & human taste▁▇▄▃▅▅█▅▄▂▁▁ #188 of 247 · 5.4% of its score · 6 tests

Beats Kev 9B v2, vLLM-SR d3-nano (2B) here, though they rank higher overall. Trails Qwen-2.5-1B-RLCD (decision readout), Bekko System One v0 68M, Crivo-17M here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: New Yorker (accuracy) submitted	52.3% ✎	▁▁▂▃▃▃▇▅█▅▃▁ beats 62/246	medium · avg 61%	▇ 2.4%	at its level
Decision Index 0.4.1: Humicroedit (accuracy) submitted	54.4% ✎	▁·▁▃▃▂▅█▄▂▁▁ beats 53/246	medium · avg 60%	▇ 1.1%	blunt · at its level
Decision Index 0.4.1: BPoMP (accuracy) submitted	65.1% ✎	▁█▄▄▅█▆▅▄▁▁▁ beats 59/246	easy · avg 83%	▇ 0.9%	sharp · noisy
Decision Index 0.4.1: cfcolor (accuracy) submitted	57.9% ✎	▁········▁▇█ beats 77/246	medium · avg 60%	▇ 0.6%	blunt · at its level
Decision Index 0.4.1: POP909-CL (accuracy) submitted	1.1% ✎	▄▅▅▃▆▅▃▅▂▅▃█ beats 19/237	hard · avg 21%	▇ 0.3%	sharp · noisy
Decision Index 0.4.1: Habermas Machine (accuracy) submitted	42.0% ✎	▁▂▄█▇▅▁▁▄▄▁▁ beats 93/246	medium · avg 47%	▇ 0.2%	blunt · at its level

knowledge & reasoning▁▂▂▂▂▃▃▅▅▆█▂ #159 of 263 · 32% of its score · 21 tests

Beats StartLux-Decision v1.1 2B, Solomon v1.1, Standard One 3B SH (v4) here, though they rank higher overall. Trails Saracura PT-BR 4B, Ryotide Qwen v0.3.1 (decision readout), Interfaze lev Qwen3.5-4B LoRA here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: MMLU-Pro (accuracy) submitted	50.7% ✎	▄▄▃▃▂▄▅▇██▅▁ beats 114/246	medium · avg 46%	▇▇ 5.6%	sharp · at its level
Decision Index 0.4.1: MMLU (accuracy) submitted	71.0% ✎	▁▄▄▃▃▂▅▇▅█▇▁ beats 98/246	easy · avg 70%	▇▇ 4.6%	sharp · at its level
Decision Index 0.4.1: BBH (accuracy) submitted	62.5% ✎	▄▆▄▄█▇▇▄▃▁▁▁ beats 92/246	easy · avg 66%	▇▇ 3.9%	at its level
Decision Index 0.4.1: ARC-Challenge (accuracy) submitted	90.3% ✎	▂▂▂▂▂▂▁▃▄▆█▆ beats 97/246	very easy · avg 90%	▇ 3.6%	sharp
Decision Index 0.4.1: ARC-Easy (accuracy) submitted	97.2% ✎	▁▁▂▁▂▂▂▂▃▆█▅ beats 105/246	very easy · avg 96%	▇ 2.2%	sharp · maxed out · noisy
Decision Index 0.4.1: WinoGrande (accuracy) submitted	70.2% ✎	█▅█▅▄█▆▄▃▃▁▁ beats 85/246	easy · avg 78%	▇ 2.1%	typical
Decision Index 0.4.1: GSM8K (accuracy (4- and 10-choice)) submitted	46.5% ✎	▁··▄█▆▆▆▅▅▃▁ beats 131/246	medium · avg 45%	▇ 1.9%	sharp · at its level
Decision Index 0.4.1: GPQA Diamond (accuracy) submitted	37.8% ✎	▂▆▅▅▆█▆▃▁··▁ beats 100/246	medium · avg 39%	▇ 1.7%	blunt · at its level

13 more tests

test	result	vs the others in its run	difficulty	counts	why
Ollaya · PubMedQA	71.6% ✎	▃▂▃▃·▂▃▃▆█▆▂ beats 15/25	easy · avg 69%	▇ 1.3%	at its level
Ollaya · BoolQ	85.3% ✎	▁··▃▃▁▄▁▃█▇▄ beats 13/25	easy · avg 84%	▇ 1.1%	typical
Decision Index 0.4.1: MuSR (accuracy) submitted	56.3% ✎	▁▁▂█▅▄▁▁▁▁▁▁ beats 87/246	medium · avg 60%	▇ 0.7%	blunt · at its level
Decision Index 0.4.1: SATA-Bench (case exact accuracy) submitted	28.8% ✎	▁▁▁▁▁▁▁▂▅█▂▁ beats 131/246	hard · avg 22%	▇ 0.6%	typical
Decision Index 0.4.1: CRUXEval (accuracy) submitted	48.1% ✎	██▄▃▃▄▁▁···▁ beats 100/246	medium · avg 58%	▇ 0.6%	at its level · noisy
Decision Index 0.4.1: CLadder (accuracy) submitted	63.1% ✎	▁███▆▅▃▃▁▁▁▁ beats 77/246	easy · avg 71%	▇ 0.6%	at its level
Ollaya · SQuAD 2 answerability	78.3% ✎	▁▁▃▁▇█▃▃▄▃·▁ beats 15/25	easy · avg 80%	▇ 0.5%	blunt
Ollaya · VitaminC verification	72.1% ✎	▁···▃·▄█▃··▁ beats 5/25	easy · avg 78%	▇ 0.5%	blunt
Ollaya · MultiNLI	91.6% ✎	▂··▄██▂▆▆█▆▂ beats 23/25	very easy · avg 87%	▇ 0.3%	blunt
Decision Index 0.4.1: ChessBench (accuracy) submitted	8.9% ✎	▄▇▄▄▄▆▄█▄▁▁▁ beats 44/246	hard · avg 16%	▇ 0.3%	blunt
Decision Index 0.4.1: HLE (accuracy) submitted	10.2% ✎	▁▁▂▃▆█▇▅▅▃▁▁ beats 32/246	very hard · avg 13%	▇ 0.0%	blunt
Hard-Decisions · ProofWriter CWA	58.8% ✎	█··········█ beats 0/1		—	not counted yet: no other model in the same run to compare it with
Hard-Decisions · ProofWriter OWA	53.6% ✎	█··········█ beats 0/1		—	not counted yet: no other model in the same run to compare it with

workflows▁▁▁▂▁▂▂▆█▄▁▁ #64 of 114 · 17% of its score · 21 tests

Beats Clef-omni, Clef-flash, Bespoke Nimble 9B v3 here, though they rank higher overall. Trails InternLM Intern-Decision Qwen3.5-4B, StartLux-Decision v1.1 0.8B here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
JevalKit: Procedural Typed Decisions	63.1% ✎	█·███▄·▄▄▄·▄ beats 4/12	easy · avg 67%	▇▇ 3.9%	sharp · at its level
Ollaya · Typed decisions test	66.9% ✎	▄·▂▆▄▄▂·▄█▆▆ beats 14/22	medium · avg 64%	▇ 2.1%	at its level
System One Bench: Typed decisions (English)	66.3% ✎	▁▁·▁▁▁▅▄▇██▁ beats 9/29	medium · avg 63%	▇ 1.9%	blunt · at its level
JevalKit: DecisionBench standard eval	74.7% ✎	▃▃·▃···▅█▅·█ beats 6/12	easy · avg 70%	▇ 1.6%	at its level
Ollaya · PAWS paraphrase	81.6% ✎	▂▂▂▂·▃█▆·▂·▁ beats 15/25	easy · avg 81%	▇ 1.3%	typical
Ollaya · MASSIVE German intent	86.0% ✎	▁····▁·▂▃▆█▃ beats 18/23	easy · avg 84%	▇ 1.2%	sharp
Ollaya · SummEval relevance	37.1% ✎	▁··▁▂▁▅··▃██ beats 9/25	medium · avg 41%	▇ 1.1%	at its level
Ollaya · Aegis 2 safety	81.2% ✎	▂·▁▁▂·▃▃▂█▃▁ beats 18/25	easy · avg 77%	▇ 1.1%	typical

13 more tests

test	result	vs the others in its run	difficulty	counts	why
JevalKit: JevBench public 231	74.0% ✎	█▄▄▄▄▄██▄··▄ beats 1/12	easy · avg 82%	▇ 1.0%	typical
Ollaya · SummEval consistency	81.3% ✎	▁·▁▁··▁▁▁▁▄█ beats 14/25	easy · avg 75%	▇ 0.7%	sharp · at its level · noisy
Ollaya · MASSIVE English intent	86.3% ✎	▁·······▂▃█▃ beats 17/23	very easy · avg 86%	▇ 0.6%	noisy
Ollaya · HelpSteer 2 scoring	34.1% ✎	▁▁··▁·▁▃▇█▅▄ beats 6/23	medium · avg 38%	▇ 0.4%	blunt · at its level
JevalKit: Typed Decisions held-out modal labels	66.2% ✎	▃▅··▃·▃▃·█▃█ beats 2/12	easy · avg 74%	▇ 0.1%	blunt · at its level
Ollaya · Civil Comments	81.7% ✎	▂▂▂·▂▅▅▆▃▅·█ beats 5/23	very easy · avg 86%	▇ 0.1%	blunt · noisy
Local Jev · English support tickets	88.6% ✎	no other model yet		—	not counted yet: no other model in the same run to compare it with
Local Jev · Korean support tickets	86.7% ✎	no other model yet		—	not counted yet: no other model in the same run to compare it with
Local Jev · Transfer v4	80.4% ✎	no other model yet		—	not counted yet: no other model in the same run to compare it with
Local Jev · Typed Decisions	67.0% ✎	no other model yet		—	not counted yet: no other model in the same run to compare it with
Local Jev · NSMC Korean sentiment	83.3% ✎	no other model yet		—	not counted yet: no other model in the same run to compare it with
Local Jev · KLUE-YNAT Korean headlines	74.0% ✎	no other model yet		—	not counted yet: no other model in the same run to compare it with
Local Jev · CLINC intent and out-of-scope	84.0% ✎	no other model yet		—	not counted yet: no other model in the same run to compare it with

other tests5 not in a task · 7.5% of its score

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: private same skills (equated points)	35.5400 points ✎	▂▂▂▃▃▃▄▅▅▇█▃ beats 79/247	avg 41.206 points	▇ 2.9%	blunt
JevalKit: JevAlt-Bench EN/TR/DE	86.5% ✎	▅··▃▃▅▃█··▅▃ beats 3/12	very easy · avg 87%	▇ 2.3%	typical
Decision Index 0.4.1: private new domains (equated points)	34.0100 points ✎	▃█▃▃▃▆▃▃▄▄▄▄ beats 102/247	avg 38.861 points	▇ 1.6%	blunt
JevalKit: Decision Index public 0.3 (official weighted points)	37.1600 points ✎	▃▃·▃·▃▃▃▅▅·█ beats 3/12	avg 39.649 points	▇ 0.7%	blunt
JevalKit: Chinese decision validation v0.2	87.2% ✎	▃····█·█▃▃█▃ beats 1/12	very easy · avg 88%	▇ 0.1%	blunt

$ where to run it

Other ways to run it

Run it as published · needs 11 GB, NVIDIA only · from github.com/ollaya-dev
ollaya run kev:4b --preset triage "The customer requests a refund."

Download model files ↗
Published model files verified. Includes adapters or readout weights; follow the model documentation for the required base model. This is a verified repository snapshot; the benchmark does not establish an identical weight revision. A matching Ollaya runtime manifest and published short-workload GPU memory measurement were reviewed separately; see local variants and identity_note. The command was not executed in this review. Source · checked 2026-10-31.

Model repository and documentation: huggingface.co/jaredpalmer ↗

$ how the score was worked out · Decision Index 0.4.1 62%, System One Bench (Ali Alhashi… 16%, Ollaya · original runs 12% and 1 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
Decision Index 0.4.1 · submitted; independent execution unverified	42	▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 62%	CC BY 4.0
System One Bench (Ali Alhashimi)	16	▇▇▇▇▇ 16%	CC-BY-4.0
Ollaya · original runs	14	▇▇▇▇ 12%	Apache-2.0
JevalKit (Valen Team)	7	▇▇▇ 10%	Apache-2.0
prior · pull toward the middle		▇ 0%
weighted average		= 48.5

not counted (11)

Decision Index 0.4.1	Decision Index 0.4.1: RouterBench (selected quality (quality objective)) [correct]	80.0%	failed run: Publisher excludes this protocol because its prompt reveals GPT-4 as the best route. Evidence: https://huggingface.co/spaces/multimodalart/jev-decision-index/blob/c130faf65f82e81d3bdefb01e758c712dc5c5199/data/index.json
Decision Index 0.4.1	Decision Index 0.4.1: SGD/SGD-X (macro-F1) [correct]	47.8%	failed run: Publisher reports a builder bug exposing dialogue turns after the classified turn; awaiting corrected rerun. Evidence: https://huggingface.co/spaces/multimodalart/jev-decision-index/blob/c130faf65f82e81d3bdefb01e758c712dc5c5199/data/index.json
Anthus · Hard-Decisions	Hard-Decisions · ProofWriter CWA [correct]	58.8%	fewer than three models have this test yet, or this source ran it for this model alone
Anthus · Hard-Decisions	Hard-Decisions · ProofWriter OWA [correct]	53.6%	fewer than three models have this test yet, or this source ran it for this model alone
tak-bro · Local Jev Bench	Local Jev · English support tickets [correct]	88.6%	fewer than three models have this test yet, or this source ran it for this model alone
tak-bro · Local Jev Bench	Local Jev · Korean support tickets [correct]	86.7%	fewer than three models have this test yet, or this source ran it for this model alone
tak-bro · Local Jev Bench	Local Jev · Transfer v4 [correct]	80.4%	fewer than three models have this test yet, or this source ran it for this model alone
tak-bro · Local Jev Bench	Local Jev · Typed Decisions [correct]	67.0%	fewer than three models have this test yet, or this source ran it for this model alone
tak-bro · Local Jev Bench	Local Jev · NSMC Korean sentiment [correct]	83.3%	fewer than three models have this test yet, or this source ran it for this model alone
tak-bro · Local Jev Bench	Local Jev · KLUE-YNAT Korean headlines [correct]	74.0%	fewer than three models have this test yet, or this source ran it for this model alone
tak-bro · Local Jev Bench	Local Jev · CLINC intent and out-of-scope [correct]	84.0%	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