whatmodel.sh / decision / vLLM-SR d3-ultra (Decision 3.0, 27B)

vLLM-SR d3-ultra (Decision 3.0, 27B)

+ contribute

vLLM-SR d3-ultra (Decision 3.0, 27B) is an open-weights decision model from vLLM Semantic Router. Its published size is 26.09B parameters. Licence: apache-2.0.

Catalogue identity notes

Exact Decision Index tested engine vllm-sr-d3-ultra; reported model vLLM-SR d3-ultra (Decision 3.0, 27B). Type: full fine-tune; technique: autoregressive. Published checkpoint/base reference: 37fee01926f3de396620e794201b074fb427ac2d. 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

· Weights licence ↗ · Parameter count ↗

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

72.3±1.3#8 of 355 decision modelsfrom 56 benchmark results across 3 sources

$ how it compares

The decision models ranked either side of vLLM-SR d3-ultra (Decision 3.0, 27B), out of 357 with a score, then more from vLLM Semantic Router.

#	model	score	where	RAM	released
6	Ainvest-Decision-27B v1.0 Ainvest-World	73±2	model files		2026-10-27
7	StartLux-Decision v1.1 27B StartLux	73±1	model files
8	vLLM-SR d3-ultra (Decision 3.0, 27B) vLLM Semantic Router	72±1	model files
9	vLLM-SR d3-turbo v1.1 (27B) vLLM Semantic Router	72±1	model files
10	Perplexity Decider v1.1 (27B) Perplexity	72±1	api price unknown		2026-10-05
	more from vLLM Semantic Router
25	vLLM-SR d3 (27B) vLLM Semantic Router	69±1	model files
97	vLLM-SR Decision 2.0 Vega 27B vLLM Semantic Router	61±1	model files
103	vLLM-SR d3-flash (9B) vLLM Semantic Router	61±1	local	~18GB

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

$ what it's good at

tools & automation▁▃▁▁▁▂▃▄▆██▆ #54 of 250 · 3.4% of its score · 6 tests

Trails baselabs joey-1 (2B), Solomon v1.1, Amber Hybrid 4B (decision readout) here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: API-Bank (accuracy) submitted	86.6% ✎	▁▁▁▁▁▁▁▂▃▂█▂ beats 202/236	easy · avg 70%	▇ 1.0%	sharp · noisy
Decision Index 0.4.1: ToolRet (nDCG@10) submitted	71.4% ✎	▁▁▁▁▁·▁▁▁▄█▃ beats 241/246	medium · avg 60%	▇ 0.8%	blunt · at its level
Decision Index 0.4.1: When2Call (accuracy) submitted	90.8% ✎	▁▁▁▂▃▅▅▆█▇█▃ beats 229/246	easy · avg 72%	▇ 0.6%	noisy
Decision Index 0.4.1: Home appliance simulator (case exact accuracy) submitted	100.0% ✎	█▁▁▃▄▄▆▄▁▂▁▃ beats 227/243	medium · avg 37%	▇ 0.5%	sharp · noisy
Decision Index 0.4.1: BFCL (case exact accuracy) submitted	97.3% ✎	▁▂▁▁▁▂▂▄▅▆█▁ beats 188/246	very easy · avg 92%	▇ 0.5%	sharp · maxed out · noisy
Decision Index 0.4.1: RouterBench (selected quality (quality objective)) submitted	79.5% ✎	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▁▁▁·▁▁▁▁▃▃█▅ #19 of 258 · 5.8% of its score · 7 tests

Beats Fastino GLiDE 1.2, KnowLine 31B Gen1, Ainvest-Decision-27B v1.0 here, though they rank higher overall. Trails JEV-35B, StartLux-Decision v1.1 4B, VisionJEV 1.0.0 (27B) here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: BRIGHT (nDCG@10) submitted	51.5% ✎	▁▁▁▁▁▁▁▂▂▄▇█ beats 245/246	medium · avg 38%	▇ 2.5%	blunt · at its level
Decision Index 0.4.1: Amazon ESCI (macro-F1) submitted	61.6% ✎	▁▁▁▁▁▁▂▂▃▃█▅ beats 227/246	medium · avg 47%	▇ 2.1%	at its level
Decision Index 0.4.1: HoVer (accuracy) submitted	89.6% ✎	▆▆▇▅▄▄▃▇▄█▆▄ beats 240/246	easy · avg 74%	▇ 0.4%	blunt
Decision Index 0.4.1: PhishNChips phishing decisions (accuracy) submitted	80.2% ✎	█▅▅▄▃▂▁▁▁▁·▁ beats 173/245	easy · avg 70%	▇ 0.4%	noisy
Decision Index 0.4.1: BANKING77 (macro-F1) submitted	94.2% ✎	▁·▁▁▁▁▁▁▂█▇█ beats 237/238	easy · avg 79%	▇ 0.2%	noisy
Decision Index 0.4.1: CLINC150+OOS (macro-F1) submitted	96.1% ✎	▁▁▁▁▁▁▁▂▃▆█▄ beats 222/238	easy · avg 83%	▇ 0.2%	sharp · maxed out · noisy
Decision Index 0.4.1: SGD/SGD-X (macro-F1) submitted	42.8% ✎	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▂▂▂▃▂▅▆▃█▆▄▃ #5 of 256 · 14% of its score · 9 tests

Beats Jeffan-31b-1710, Sieve 31B v3, Fastino GLiDE 1.2 here, though they rank higher overall. Trails AutoTrust-JEV-27B v1.1, vLLM-SR d3-turbo v1.1 (27B) here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: ANLI (macro-F1) submitted	76.5% ✎	▁▂▃▅▅█▂····▁ beats 211/246	medium · avg 60%	▇ 3.7%	typical
Decision Index 0.4.1: NLI4CT (macro-F1) submitted	86.3% ✎	▁▁▁▂▂▃▃▄▇▇█▄ beats 229/246	easy · avg 74%	▇ 3.4%	typical
Decision Index 0.4.1: iSarcasmEval (Sarcasm F1 · track A, English) submitted	54.7% ✎	▁·▁·▁▁▂▄▅█▇▄ beats 157/246	medium · avg 44%	▇ 2.0%	at its level
Decision Index 0.4.1: ACOS (per-review F1) submitted	63.6% ✎	▁··▂▃▃▅▇█▇▃▃ beats 240/246	hard · avg 22%	▇ 1.5%	at its level · noisy
Decision Index 0.4.1: HellaSwag (accuracy) submitted	98.2% ✎	▂▄▂▂▄▇▃▇█▅▃▁ beats 239/246	very easy · avg 85%	▇ 1.5%	sharp · maxed out
Decision Index 0.4.1: VAST (macro-F1) submitted	82.9% ✎	▁▁▁▂▅▄▄▂▄▆█▄ beats 245/246	medium · avg 62%	▇ 0.8%	typical
Decision Index 0.4.1: ContractNLI (macro-F1) submitted	84.7% ✎	▁▁▁▁▁▁▁▂▂▇▇█ beats 221/246	easy · avg 71%	▇ 0.7%	noisy
Decision Index 0.4.1: RAGTruth (F1 on hallucinated class) submitted	83.4% ✎	▁▁▁▁▁▁▁▁▁▃▆█ beats 213/246	easy · avg 66%	▇ 0.5%	noisy

1 more test

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: FinEntity (macro-F1) submitted	93.3% ✎	▁···▁▁▁▁▃▆█▃ beats 200/246	easy · avg 84%	▇ 0.3%	maxed out · noisy

arts & human taste▁▇▄▃▅▅█▅▄▂▃▁ #11 of 247 · 7.3% of its score · 6 tests

Beats Jeffan-31b-1710, KnowLine 31B Gen1, StartLux-Decision v1.1 27B here, though they rank higher overall. Trails AnyJev Tacitus 31B, baselabs elephant-1 (9B), vera-large 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	84.1% ✎	▁▁▂▃▃▃▇▅█▅▃▁ beats 239/246	medium · avg 61%	▇ 2.7%	typical
Decision Index 0.4.1: Humicroedit (accuracy) submitted	67.7% ✎	▁·▁▃▂▂▅█▄▃▁▁ beats 229/246	medium · avg 60%	▇ 1.8%	blunt · at its level
Decision Index 0.4.1: cfcolor (accuracy) submitted	70.3% ✎	▁········▁▇█ beats 243/246	medium · avg 60%	▇ 1.0%	blunt · at its level
Decision Index 0.4.1: POP909-CL (accuracy) submitted	78.3% ✎	▄▅▅▃▆▅▃▅▂▅▃█ beats 209/237	hard · avg 21%	▇ 0.9%	sharp · at its level · noisy
Decision Index 0.4.1: BPoMP (accuracy) submitted	98.3% ✎	▁█▄▄▅█▆▅▄▃▁▁ beats 236/246	easy · avg 83%	▇ 0.5%	sharp · maxed out · noisy
Decision Index 0.4.1: Habermas Machine (accuracy) submitted	62.5% ✎	▁▂▄█▇▅▁▁▄▄▁▁ beats 225/246	medium · avg 47%	▇ 0.3%	blunt · at its level

knowledge & reasoning▁▂▂▂▂▃▃▅▅▆█▃ #6 of 263 · 39% of its score · 15 tests

Beats Fastino GLiDE 1.2, KnowLine 31B Gen1, Ainvest-Decision-27B v1.0 here, though they rank higher overall. Trails Jev 1.13, Perplexity Decider v1.1 (27B) here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
Etsabary · 1342 text decisions	83.8% ✎	▁··▂·▃▅▁▃▂▂█ beats 21/24	medium · avg 47%	▇▇▇▇ 11%	sharp
Decision Index 0.4.1: MMLU-Pro (accuracy) submitted	76.0% ✎	▄▄▃▃▂▄▅▇██▅▁ beats 241/246	medium · avg 46%	▇▇▇ 7.6%	sharp
Decision Index 0.4.1: BBH (accuracy) submitted	92.4% ✎	▄▆▄▄█▇▇▄▃▃▁▁ beats 239/246	easy · avg 66%	▇ 3.7%	typical
Decision Index 0.4.1: MMLU (accuracy) submitted	88.7% ✎	▁▄▄▃▃▂▅▇▅█▇▁ beats 236/246	easy · avg 70%	▇ 3.5%	sharp
Decision Index 0.4.1: GPQA Diamond (accuracy) submitted	60.7% ✎	▂▆▅▅▆█▆▃▃··▁ beats 245/246	medium · avg 39%	▇ 3.2%	blunt · at its level
Decision Index 0.4.1: GSM8K (accuracy (4- and 10-choice)) submitted	85.3% ✎	▁··▄█▆▆▆▅▅▃▁ beats 234/246	medium · avg 45%	▇ 2.4%	sharp
Decision Index 0.4.1: SATA-Bench (case exact accuracy) submitted	38.6% ✎	▁▁▁▁▁▁▁▂▅█▃▁ beats 219/246	hard · avg 22%	▇ 1.8%	at its level
Decision Index 0.4.1: WinoGrande (accuracy) submitted	96.0% ✎	█▅█▅▄█▆▄▃▃▃▁ beats 241/246	easy · avg 78%	▇ 1.7%	typical

7 more tests

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: MuSR (accuracy) submitted	79.9% ✎	▁▁▂█▅▄▁▃▁▁▁▁ beats 233/246	medium · avg 60%	▇ 1.1%	blunt · at its level
Decision Index 0.4.1: ChessBench (accuracy) submitted	26.6% ✎	▄▇▄▄▄▆▄█▄▁▁▁ beats 230/246	hard · avg 16%	▇ 0.9%	blunt
Decision Index 0.4.1: ARC-Challenge (accuracy) submitted	97.5% ✎	▂▂▂▂▂▂▁▃▄▆█▆ beats 226/246	very easy · avg 90%	▇ 0.8%	sharp · maxed out
Decision Index 0.4.1: CRUXEval (accuracy) submitted	89.5% ✎	██▄▃▃▄▁▁···▁ beats 239/246	medium · avg 58%	▇ 0.8%	noisy
Decision Index 0.4.1: CLadder (accuracy) submitted	94.9% ✎	▁███▆▅▃▃▁▃▁▁ beats 238/246	easy · avg 71%	▇ 0.7%	typical
Decision Index 0.4.1: ARC-Easy (accuracy) submitted	99.3% ✎	▁▁▂▁▂▂▂▂▃▆█▅ beats 231/246	very easy · avg 96%	▇ 0.3%	sharp · maxed out · noisy
Decision Index 0.4.1: HLE (accuracy) submitted	19.6% ✎	▁▁▂▃▆█▇▅▅▃▁▃ beats 243/246	very hard · avg 13%	▇ 0.0%	blunt

vision▁·▃▂▃▃▇▄▄█▄▃ #10 of 79 · 21% of its score · 13 tests

Beats Fastino GLiDE 1.2, StartLux-Decision v1.1 27B here, though they rank higher overall. Trails Mercury Decide (free endpoint), Clef, Jev 1.14 here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
Etsabary · 200 image decisions	74.0% ✎	▂▂··▂▃▂·█▃▂▂ beats 11/14	medium · avg 45%	▇▇▇ 8.3%	sharp
Decision Index Vision 0.4 · Mind2Web (chance-corrected points) submitted	85.8900 points ✎	▁·▁▁▃▁▄▄▃▅█▂ beats 55/61	avg 63.423 points	▇ 1.6%	blunt
Decision Index Vision 0.4 · MMMU-Pro vision (chance-corrected points) submitted	43.6500 points ✎	▁▁▅▁▄▁▄█▄▃▇▇ beats 49/61	avg 22.797 points	▇ 1.5%	blunt
Decision Index Vision 0.4 · Winoground (chance-corrected points) submitted	84.0000 points ✎	▁·▁▁▁▁·▂▁▃▅█ beats 44/61	avg 66.987 points	▇ 1.3%	blunt
Decision Index Vision 0.4: private same skills (points)	65.7300 points ✎	▁▁▃▃▁▂▄▆▃▃▅█ beats 48/61	avg 48.067 points	▇ 1.3%	blunt
Decision Index Vision 0.4 · Moderation (Hateful Memes) (chance-corrected points) submitted	41.3000 points ✎	▆▃▃▆▄▄▇██▁▆▆ beats 39/61	avg 28.443 points	▇ 1.2%	blunt
Decision Index Vision 0.4 · RealWorldQA (chance-corrected points) submitted	69.4800 points ✎	▁·▂▂▁▃▃▅▅▃▃█ beats 44/61	avg 53.651 points	▇ 1.1%	blunt
Decision Index Vision 0.4 · CharXiv (chance-corrected points) submitted	80.4500 points ✎	▂▁▁▁▁▃▁▅▅▆██ beats 53/61	avg 62.137 points	▇ 1.0%	blunt

5 more tests

test	result	vs the others in its run	difficulty	counts	why
Decision Index Vision 0.4 · InfographicVQA (chance-corrected points) submitted	96.7200 points ✎	▁·▁▂▁▁▃▁▂▂▅█ beats 45/61	avg 83.859 points	▇ 0.8%	blunt
Decision Index Vision 0.4 · R-Bench-M (chance-corrected points) submitted	43.1500 points ✎	▂▄▇▆█▄▄▇▃··▁ beats 60/61	avg 19.676 points	▇ 0.8%	blunt
Decision Index Vision 0.4 · BLINK (chance-corrected points) submitted	71.3300 points ✎	▁···▁▂▂▂█▄▃▃ beats 53/61	avg 44.994 points	▇ 0.8%	blunt
Decision Index Vision 0.4 · CV-Bench (chance-corrected points) submitted	81.7900 points ✎	▁▁▁▁▂▁▁▁▄█▃▃ beats 60/61	avg 67.662 points	▇ 0.6%	blunt
Decision Index Vision 0.4 · KIE (CORD+FUNSD) (chance-corrected points) submitted	92.2500 points ✎	▁··▁▁▁▁▁▂▄█▁ beats 41/61	avg 85.184 points	▇ 0.4%	blunt

other tests2 not in a task · 8.4% of its score

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: private same skills (equated points)	66.4900 points ✎	▂▂▂▃▃▃▄▅▅▇█▃ beats 244/247	avg 41.206 points	▇▇ 5.4%	blunt
Decision Index 0.4.1: private new domains (equated points)	75.7100 points ✎	▃█▃▃▂▆▃▃▄▄▄▄ beats 242/247	avg 38.861 points	▇ 3.0%	blunt

$ where to run it

Download model files ↗
Published model files verified. This is the revision recorded by the benchmark. 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/vllm-sr ↗

$ how the score was worked out · Decision Index 0.4.1 68%, Etsabary · decision benchmarks 19%, Decision Index Vision 0.4 12%

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	▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 68%	CC BY 4.0
Etsabary · decision benchmarks	2	▇▇▇▇▇▇ 19%	MIT
Decision Index Vision 0.4 · submitted; independent execution unverified	12	▇▇▇▇ 12%	CC BY 4.0
prior · pull toward the middle		▇ 1%
weighted average		= 72.3

not counted (2)

Decision Index 0.4.1	Decision Index 0.4.1: RouterBench (selected quality (quality objective)) [correct]	79.5%	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]	42.8%	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

$ 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