whatmodel.sh / decision / Together Tev1 Qwen3.5-4B

Together Tev1 Qwen3.5-4B

+ contribute

Together Tev1 Qwen3.5-4B is an open-weights decision model from Together. Its published size is 4.66B parameters.

Catalogue identity notes

Exact Decision Index tested engine tev1-4b; reported model Together Tev1 Qwen3.5-4B. Type: full fine-tune; technique: autoregressive. Implementation: https://github.com/togethercomputer/tev1. Comparison entry; no current run command, 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.

Model information sources

· Parameter count ↗

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

50.6±1.0#200 of 355 decision modelsfrom 61 benchmark results across 3 sources

$ how it compares

The decision models ranked either side of Together Tev1 Qwen3.5-4B, out of 357 with a score, then more from Together.

#	model	score	where	RAM
200	vLLM-SR Decision 2.0 Nox 4B vLLM Semantic Router	51±1	local	~12GB
201	Decider 4B Mapika	51±1	local	~11GB
202	Together Tev1 Qwen3.5-4B Together	51±1	model files
203	OneJev-4B (JevBench mini) OmniJev	51±2	source code
204	JPT-4B Qwen3.5-4B LoRA kirp	50±1	model files
	more from Together
154	Tev1-4B-experimental (DecisionBench 0b7becf0) Together	56±2	model files
306	Together Tev1 Qwen3.5-0.8B Together	31±1	model files

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

$ what it's good at

tools & automation▁▃▁▁▁▂▃▄▆██▆ #175 of 250 · 8.5% of its score · 7 tests

Beats LiquidAI d1-3B, Decider 4B here, though they rank higher overall. Trails JevK5 Lite, Mola-150M, Gevva E2B here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
DecideBench v1: agent routing	100.0% ✎	▄·····▂▄·█·▄ beats 9/10	very easy · avg 93%	▇ 3.1%	sharp · maxed out · noisy
DecideBench v1: action review	86.0% ✎	▃▃·▃·█···█·▅ beats 3/10	easy · avg 87%	▇ 1.7%	sharp · noisy
Decision Index 0.4.1: BFCL (case exact accuracy) submitted	90.2% ✎	▁▂▁▁▁▂▂▄▅▆█▁ beats 74/246	very easy · avg 92%	▇ 1.6%	sharp · maxed out · noisy
Decision Index 0.4.1: When2Call (accuracy) submitted	60.5% ✎	▁▁▁▂▃▅▅▆█▇█▃ beats 67/246	easy · avg 72%	▇ 0.7%	at its level · noisy
Decision Index 0.4.1: Home appliance simulator (case exact accuracy) submitted	14.8% ✎	█▁▁▃▄▄▆▄▁▂▁▃ beats 78/243	medium · avg 37%	▇ 0.7%	sharp · at its level · noisy
Decision Index 0.4.1: ToolRet (nDCG@10) submitted	65.0% ✎	▁▁▁▁▁·▁▁▁▄█▁ beats 134/246	medium · avg 60%	▇ 0.6%	blunt · at its level
Decision Index 0.4.1: RouterBench (selected quality (quality objective)) submitted	67.8% ✎	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▁▁▁·▁▁▁▁▃▃█▅ #176 of 258 · 13% of its score · 12 tests

Beats Solar Decide, JevK5 9B v0.3.3, Aplomb 1 here, though they rank higher overall. Trails Mola-68M, Bosun v3.1 1.7B LoRA, Dinah-1 here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
System One Bench: MIRACL pairwise (Arabic)	87.5% ✎	▂···▁··▃▄▃█▇ beats 12/29	easy · avg 84%	▇ 2.7%	at its level
System One Bench: MASSIVE intent (Arabic)	70.0% ✎	▄·▁·▃▃▃▃█▄▇▇ beats 4/29	easy · avg 76%	▇ 2.6%	typical
System One Bench: MIRACL rerank (Arabic)	85.5% ✎	▁▂·▁▁·▂·▄▄█▇ beats 11/29	easy · avg 79%	▇ 1.6%	typical
Decision Index 0.4.1: BRIGHT (nDCG@10) submitted	43.0% ✎	▁▁▁▁▁▁▁▂▂▄▇█ beats 129/246	medium · avg 38%	▇ 1.4%	blunt · at its level
Decision Index 0.4.1: Amazon ESCI (macro-F1) submitted	44.2% ✎	▁▁▁▁▁▁▂▂▃▃█▅ beats 79/246	medium · avg 47%	▇ 1.3%	at its level
DecideBench v1: support intent	100.0% ✎	▃····▂▂·▃··█ beats 7/10	very easy · avg 95%	▇ 0.9%	sharp · maxed out · noisy
System One Bench: MASSIVE scenario (Arabic)	56.5% ✎	▃▁·▄▃▂█▂▁▁·▃ beats 5/29	medium · avg 60%	▇ 0.9%	at its level
Decision Index 0.4.1: HoVer (accuracy) submitted	64.1% ✎	▆▆▇▅▄▄▃▇▄█▆▄ beats 74/246	easy · avg 74%	▇ 0.4%	blunt · at its level

4 more tests

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: PhishNChips phishing decisions (accuracy) submitted	50.6% ✎	█▅▅▄▃▂▁▁▁▁·▁ beats 41/245	easy · avg 70%	▇ 0.4%	at its level · noisy
System One Bench: AG News (English)	91.5% ✎	▁···▅▁▅▂▆█▂▅ beats 26/29	easy · avg 88%	▇ 0.2%	blunt
System One Bench: Banking77 (English)	1.0% ✎	█▅▆█▅▃▃▃···▂ beats 0/26	very easy · avg 87%	—	failed run: All requests failed or were unsupported in this evaluator configuration; not a measurement of ability.
Decision Index 0.4.1: SGD/SGD-X (macro-F1) submitted	40.7% ✎	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▂▂▂▃▂▅▆▃█▆▄▃ #158 of 256 · 30% of its score · 21 tests

Beats PurpleMIST-Flash-1.0, Hopper 12B (trained), Jev-Omni here, though they rank higher overall. Trails DecisionTune 1.0, Decider 2B, Crivo-1B 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	81.9% ✎	▂▄▂▂▄▇▃▇█▅▃▁ beats 93/246	very easy · avg 85%	▇▇ 4.2%	sharp
DecideBench v1: review sentiment	92.0% ✎	▂··▂···▂▂█▄▂ beats 4/10	easy · avg 85%	▇▇ 4.2%	sharp
Decision Index 0.4.1: NLI4CT (macro-F1) submitted	77.3% ✎	▁▁▁▂▂▃▃▄▇▇█▄ beats 121/246	easy · avg 74%	▇ 3.6%	at its level
Decision Index 0.4.1: ANLI (macro-F1) submitted	56.4% ✎	▁▂▃▅▅█▂····▁ beats 93/246	medium · avg 60%	▇ 3.4%	at its level
DecideBench v1: content moderation	90.0% ✎	▂··▂···▄▄▂·█ beats 5/10	easy · avg 83%	▇ 2.7%	sharp · noisy
System One Bench: OSACT4 offensive (Arabic)	84.5% ✎	▁······▁▃█▇▄ beats 8/29	easy · avg 84%	▇ 1.9%	sharp · at its level
System One Bench: BoolQ (English)	91.0% ✎	▁▁▁▃·▁▃▃▇▇█▅ beats 20/29	easy · avg 86%	▇ 1.8%	at its level
Decision Index 0.4.1: iSarcasmEval (Sarcasm F1 · track A, English) submitted	35.9% ✎	▁·▁·▁▁▂▄▅█▇▄ beats 75/246	medium · avg 44%	▇ 1.4%	at its level

13 more tests

test	result	vs the others in its run	difficulty	counts	why
System One Bench: SST-2 (English)	94.0% ✎	▁·▃▁▂▂▃▅█·▂▁ beats 10/29	easy · avg 92%	▇ 1.2%	typical
DecideBench v1: claim support	90.0% ✎	▄··▄·▄·▄▄███ beats 4/10	easy · avg 82%	▇ 1.1%	sharp · noisy
System One Bench: SST-5 (English)	43.0% ✎	▃·▃▁▃·▇▄▄▄█▅ beats 2/29	medium · avg 51%	▇ 0.8%	blunt · at its level
Decision Index 0.4.1: ContractNLI (macro-F1) submitted	67.7% ✎	▁▁▁▁▁▁▁▂▃▇▇█ beats 67/246	easy · avg 71%	▇ 0.7%	at its level · noisy
Decision Index 0.4.1: VAST (macro-F1) submitted	47.7% ✎	▁▁▁▂▅▄▄▂▄▆█▄ beats 67/246	medium · avg 62%	▇ 0.6%	at its level
Decision Index 0.4.1: ACOS (per-review F1) submitted	10.3% ✎	▁··▂▃▃▅▇█▇▃▃ beats 60/246	hard · avg 22%	▇ 0.6%	noisy
Decision Index 0.4.1: RAGTruth (F1 on hallucinated class) submitted	50.3% ✎	▁▁▁▁▁▁▁▁▃▃▆█ beats 46/246	easy · avg 66%	▇ 0.5%	at its level · noisy
Decision Index 0.4.1: FinEntity (macro-F1) submitted	86.3% ✎	▁···▁▁▁▁▃▆█▃ beats 108/246	easy · avg 84%	▇ 0.5%	noisy
System One Bench: XNLI Arabic (Arabic)	76.0% ✎	▄▁▃▁▄▇█▅▃▁·▃ beats 20/29	medium · avg 68%	▇ 0.4%	at its level
System One Bench: XNLI French (French)	82.7% ✎	▂▅·▁▁▅▅█▃▁·▃ beats 19/29	medium · avg 76%	▇ 0.4%	blunt · at its level
System One Bench: Tweet sentiment (Arabic)	70.5% ✎	▁······▁▁▄█▄ beats 19/29	medium · avg 62%	▇ 0.3%	blunt · at its level
System One Bench: XNLI English (English)	87.3% ✎	▃·▁·▃▇█▄█▃▃▁ beats 14/29	easy · avg 85%	▇ 0.1%	blunt · noisy
System One Bench: DAIR Emotion (English)	56.5% ✎	██▃·····▁··▁ beats 11/29	medium · avg 59%	▇ 0.0%	blunt · at its level

arts & human taste▁▇▄▃▅▅█▅▄▂▁▁ #168 of 247 · 5.4% of its score · 5 tests

Beats PurpleMIST-Flash-1.0, vLLM-SR Decision 2.0 Nox 4B here, though they rank higher overall. Trails Crivo-17M, Crivo-32M v2, Crivo-68M 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	55.5% ✎	▁▁▂▃▃▃▇▅█▅▃▁ beats 75/246	medium · avg 61%	▇ 2.6%	at its level
Decision Index 0.4.1: Humicroedit (accuracy) submitted	58.3% ✎	▁·▁▃▂▂▅█▄▂▁▁ beats 79/246	medium · avg 60%	▇ 1.2%	blunt · at its level
Decision Index 0.4.1: BPoMP (accuracy) submitted	76.6% ✎	▁█▄▄▅█▆▅▄▁▁▁ beats 79/246	easy · avg 83%	▇ 0.9%	sharp · noisy
Decision Index 0.4.1: cfcolor (accuracy) submitted	58.4% ✎	▁········▁▇█ beats 84/246	medium · avg 60%	▇ 0.7%	blunt · at its level
Decision Index 0.4.1: Habermas Machine (accuracy) submitted	41.6% ✎	▁▂▄█▇▅▁▁▄▄▁▁ beats 89/246	medium · avg 47%	▇ 0.2%	blunt · at its level

knowledge & reasoning▁▂▂▂▂▃▃▅▅▆█▂ #157 of 263 · 30% of its score · 14 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	45.4% ✎	▄▄▃▃▂▄▅▇██▅▁ beats 99/246	medium · avg 46%	▇▇▇ 6.3%	sharp · at its level
Decision Index 0.4.1: MMLU (accuracy) submitted	72.3% ✎	▁▄▄▃▃▂▅▇▅█▇▁ beats 104/246	easy · avg 70%	▇▇ 4.8%	sharp · at its level
Decision Index 0.4.1: BBH (accuracy) submitted	63.5% ✎	▄▆▄▄█▇▇▄▃▁▁▁ beats 98/246	easy · avg 66%	▇▇ 4.2%	at its level
Decision Index 0.4.1: ARC-Challenge (accuracy) submitted	92.2% ✎	▂▂▂▂▂▂▁▃▄▆█▆ beats 108/246	very easy · avg 90%	▇ 3.4%	sharp
Decision Index 0.4.1: WinoGrande (accuracy) submitted	69.6% ✎	█▅█▅▄█▆▄▃▃▁▁ beats 80/246	easy · avg 78%	▇ 2.2%	typical
Decision Index 0.4.1: GSM8K (accuracy (4- and 10-choice)) submitted	29.5% ✎	▁··▄█▆▆▆▅▅▃▁ beats 79/246	medium · avg 45%	▇ 2.1%	sharp · at its level
Decision Index 0.4.1: ARC-Easy (accuracy) submitted	97.6% ✎	▁▁▂▁▂▂▂▂▃▆█▅ beats 112/246	very easy · avg 96%	▇ 1.9%	sharp · maxed out · noisy
Decision Index 0.4.1: GPQA Diamond (accuracy) submitted	42.4% ✎	▂▆▅▅▆█▆▃▁··▁ beats 131/246	medium · avg 39%	▇ 1.9%	blunt · at its level

6 more tests

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: MuSR (accuracy) submitted	58.0% ✎	▁▁▂█▅▄▁▁▁▁▁▁ beats 106/246	medium · avg 60%	▇ 0.8%	blunt · at its level
Decision Index 0.4.1: SATA-Bench (case exact accuracy) submitted	20.3% ✎	▁▁▁▁▁▁▁▂▅█▂▁ beats 76/246	hard · avg 22%	▇ 0.8%	typical
Decision Index 0.4.1: CRUXEval (accuracy) submitted	50.2% ✎	██▄▃▃▄▁▁···▁ beats 111/246	medium · avg 58%	▇ 0.7%	at its level · noisy
Decision Index 0.4.1: CLadder (accuracy) submitted	63.1% ✎	▁███▆▅▃▃▁▁▁▁ beats 78/246	easy · avg 71%	▇ 0.6%	at its level
Decision Index 0.4.1: ChessBench (accuracy) submitted	22.0% ✎	▄▇▄▄▄▆▄█▄▁▁▁ beats 189/246	hard · avg 16%	▇ 0.4%	blunt
Decision Index 0.4.1: HLE (accuracy) submitted	11.7% ✎	▁▁▂▃▆█▇▅▅▃▁▁ beats 93/246	very hard · avg 13%	▇ 0.0%	blunt

workflows▁▁▁▂▁▂▂▆█▄▁▁ #48 of 114 · 8.0% of its score · 3 tests

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

test	result	vs the others in its run	difficulty	counts	why
DecideBench v1: ticket triage	94.0% ✎	▄·····▂█·▆·▂ beats 3/10	very easy · avg 90%	▇▇ 4.2%	sharp
System One Bench: Typed decisions (English)	67.0% ✎	▁▁·▁▁▁▅▄▇██▁ beats 11/29	medium · avg 63%	▇ 2.1%	blunt · at its level
DecideBench v1: returns policy	90.0% ✎	▃··▃··▃·▅▅█▃ beats 5/10	easy · avg 81%	▇ 1.6%	sharp · noisy

other tests2 not in a task · 5.0% of its score

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: private same skills (equated points)	31.0900 points ✎	▂▂▂▃▃▃▄▅▅▇█▃ beats 68/247	avg 41.206 points	▇ 3.2%	blunt
Decision Index 0.4.1: private new domains (equated points)	37.6800 points ✎	▃█▃▃▂▆▃▃▄▄▄▄ beats 113/247	avg 38.861 points	▇ 1.8%	blunt

$ where to run it

Download model files ↗
Published model files verified. This is a verified repository snapshot; the benchmark does not establish an identical weight revision. A local run command and hardware requirements have not been verified for this entry. Source · checked 2026-10-31.

Model repository and documentation: huggingface.co/togethercomputer ↗

$ how the score was worked out · Decision Index 0.4.1 63%, DecideBench (Cho Yin Yong) 20%, System One Bench (Ali Alhashi… 17%

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	38	▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 63%	CC BY 4.0
DecideBench (Cho Yin Yong)	8	▇▇▇▇▇▇ 20%	CC BY 4.0
System One Bench (Ali Alhashimi)	15	▇▇▇▇▇ 17%	CC-BY-4.0
prior · pull toward the middle		▇ 0%
weighted average		= 50.6

not counted (3)

Decision Index 0.4.1	Decision Index 0.4.1: RouterBench (selected quality (quality objective)) [correct]	67.8%	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]	40.7%	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
System One Bench (Ali Alhashimi)	System One Bench: Banking77 (English) [correct]	1.0%	failed run: All requests failed or were unsupported in this evaluator configuration; not a measurement of ability.

$ 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