Sieve 27B v2

+ contribute

Sieve 27B v2 is an open-weights decision model from Sthānika AI, released 18 October 2026. Its published size is 27.78B parameters. Licence: apache-2.0.

Catalogue identity notes

Exact Decision Index tested engine sieve-27b-v2; reported model Sieve 27B v2. Type: LoRA + head; technique: autoregressive. 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

Release date ↗ · Weights licence ↗ · Parameter count ↗

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

66.9±1.4#37 of 355 decision modelsfrom 42 benchmark results across 1 source

$ how it compares

The decision models ranked either side of Sieve 27B v2, out of 357 with a score, then more from Sthānika AI.

#	model	score	where	released
35	mogan-decision-31B-it Mogan AI	67±1	model files
36	oenogarum (Gemma 4 31B) plantwaterer	67±1	model files
37	Sieve 27B v2 Sthānika AI	67±1	model files	2026-10-18
38	H2O Lightning 31B H2O.ai	67±4	request access
39	Aleria Judge 1.1.1 (Gemma 4 31B) (decision readout) ALERIA-ML	67±1	source code	2026-10-19
	more from Sthānika AI
3	Sieve 31B v3 Sthānika AI	74±2	model files	2026-10-27
79	Sieve-Vision-27B Sthānika AI	63±4	model files
137	Sieve 9B v2 Sthānika AI	57±1	model files	2026-10-13

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

$ what it's good at

tools & automation▁▃▁▁▁▂▃▄▆██▆ #93 of 250 · 5.8% of its score · 6 tests

Trails OneJev1.6-0.8B, Jev-Omni, baselabs joey-1 (2B) 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	89.0% ✎	▁▁▁▁▁▁▁▂▃▂█▃ beats 227/236	easy · avg 70%	▇ 1.7%	sharp · noisy
Decision Index 0.4.1: ToolRet (nDCG@10) submitted	68.5% ✎	▁▁▁▁▁·▁▁▁▄█▁ beats 214/246	medium · avg 60%	▇ 1.1%	blunt · at its level
Decision Index 0.4.1: Home appliance simulator (case exact accuracy) submitted	100.0% ✎	█▁▁▃▄▄▆▄▁▂▁▃ beats 236/243	medium · avg 37%	▇ 1.0%	sharp · noisy
Decision Index 0.4.1: When2Call (accuracy) submitted	84.8% ✎	▁▁▁▂▃▅▅▆█▇█▃ beats 170/246	easy · avg 72%	▇ 1.0%	noisy
Decision Index 0.4.1: BFCL (case exact accuracy) submitted	96.0% ✎	▁▂▁▁▁▂▂▄▅▆█▁ beats 152/246	very easy · avg 92%	▇ 0.9%	sharp · maxed out · noisy
Decision Index 0.4.1: RouterBench (selected quality (quality objective)) submitted	79.8% ✎	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▁▁▁·▁▁▁▁▃▃█▅ #24 of 258 · 7.8% of its score · 7 tests

Beats KnowLine 31B Gen1, Ainvest-Decision-27B v1.0, Perplexity Decider v1.1 (27B) here, though they rank higher overall. Trails JEV-35B, StartLux-Decision v1.1 4B, Torchcast Decision 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	46.1% ✎	▁▁▁▁▁▁▁▂▂▄▇█ beats 168/246	medium · avg 38%	▇ 3.2%	blunt · at its level
Decision Index 0.4.1: Amazon ESCI (macro-F1) submitted	61.3% ✎	▁▁▁▁▁▁▂▂▃▃█▅ beats 223/246	medium · avg 47%	▇ 2.8%	at its level
Decision Index 0.4.1: HoVer (accuracy) submitted	88.0% ✎	▆▆▇▅▄▄▃▇▄█▆▄ beats 216/246	easy · avg 74%	▇ 0.6%	blunt
Decision Index 0.4.1: PhishNChips phishing decisions (accuracy) submitted	96.7% ✎	█▅▅▄▃▂▁▃▁▁·▁ beats 237/245	easy · avg 70%	▇ 0.6%	noisy
Decision Index 0.4.1: BANKING77 (macro-F1) submitted	90.0% ✎	▁·▁▁▁▁▁▁▂█▇█ beats 178/238	easy · avg 79%	▇ 0.3%	noisy
Decision Index 0.4.1: CLINC150+OOS (macro-F1) submitted	90.9% ✎	▁▁▁▁▁▁▁▂▃▆█▄ beats 140/238	easy · avg 83%	▇ 0.3%	sharp · maxed out · noisy
Decision Index 0.4.1: SGD/SGD-X (macro-F1) submitted	46.5% ✎	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▂▂▂▃▂▅▆▃█▆▄▃ #25 of 256 · 22% of its score · 9 tests

Beats KnowLine 31B Gen1, Ainvest-Decision-27B v1.0, WeMM-JEV-27B here, though they rank higher overall. Trails Clef-flash, Clef-omni, JEV-35B 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	75.2% ✎	▁▂▃▅▅█▂····▁ beats 197/246	medium · avg 60%	▇▇ 5.5%	typical
Decision Index 0.4.1: NLI4CT (macro-F1) submitted	85.4% ✎	▁▁▁▂▂▃▃▄▇▇█▄ beats 219/246	easy · avg 74%	▇▇ 5.2%	typical
Decision Index 0.4.1: HellaSwag (accuracy) submitted	97.0% ✎	▂▄▂▂▄▇▃▇█▅▃▁ beats 224/246	very easy · avg 85%	▇ 2.9%	sharp · maxed out
Decision Index 0.4.1: iSarcasmEval (Sarcasm F1 · track A, English) submitted	47.3% ✎	▁·▁·▁▁▂▄▅█▇▄ beats 120/246	medium · avg 44%	▇ 2.8%	at its level
Decision Index 0.4.1: ACOS (per-review F1) submitted	48.3% ✎	▁··▂▃▃▅▇█▇▃▃ beats 220/246	hard · avg 22%	▇ 1.9%	at its level · noisy
Decision Index 0.4.1: VAST (macro-F1) submitted	78.1% ✎	▁▁▁▂▅▄▄▂▄▆█▄ beats 197/246	medium · avg 62%	▇ 1.1%	typical
Decision Index 0.4.1: ContractNLI (macro-F1) submitted	83.4% ✎	▁▁▁▁▁▁▁▂▂▇▇█ beats 201/246	easy · avg 71%	▇ 1.0%	noisy
Decision Index 0.4.1: RAGTruth (F1 on hallucinated class) submitted	74.0% ✎	▁▁▁▁▁▁▁▁▁▃▆█ beats 122/246	easy · avg 66%	▇ 0.7%	noisy

1 more test

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: FinEntity (macro-F1) submitted	93.2% ✎	▁···▁▁▁▁▃▆█▃ beats 197/246	easy · avg 84%	▇ 0.5%	maxed out · noisy

arts & human taste▁▇▄▃▅▅█▅▄▃▁▁ #14 of 247 · 10% of its score · 6 tests

Beats Jeffan-31b-1710, KnowLine 31B Gen1, vLLM-SR d3-turbo v1.1 (27B) here, though they rank higher overall. Trails AnyJev L0 · Gemma-4-31B (decision readout) 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	78.0% ✎	▁▁▂▃▃▃▇▅█▅▃▁ beats 207/246	medium · avg 61%	▇▇ 4.0%	typical
Decision Index 0.4.1: Humicroedit (accuracy) submitted	60.1% ✎	▁·▁▃▂▂▅█▄▂▁▁ beats 97/246	medium · avg 60%	▇ 2.4%	blunt · at its level
Decision Index 0.4.1: POP909-CL (accuracy) submitted	83.3% ✎	▄▅▅▃▆▅▃▅▂▅▃█ beats 233/237	hard · avg 21%	▇ 1.3%	sharp · at its level · noisy
Decision Index 0.4.1: cfcolor (accuracy) submitted	64.8% ✎	▁········▁▇█ beats 174/246	medium · avg 60%	▇ 1.3%	blunt · at its level
Decision Index 0.4.1: BPoMP (accuracy) submitted	98.0% ✎	▁█▄▄▅█▆▅▄▃▁▁ beats 233/246	easy · avg 83%	▇ 0.9%	sharp · maxed out · noisy
Decision Index 0.4.1: Habermas Machine (accuracy) submitted	65.7% ✎	▁▂▄█▇▅▁▁▄▄▃▁ beats 241/246	medium · avg 47%	▇ 0.4%	blunt · at its level

knowledge & reasoning▁▂▂▂▂▃▃▅▅▆█▂ #79 of 263 · 43% of its score · 14 tests

Beats StartLux-Decision v1.1 9B here, though they rank higher overall. Trails Solar Decide, Clef-flash, Decider chat · Qwen3.6-27B (decision readout) 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	62.4% ✎	▄▄▃▃▂▄▅▇██▅▁ beats 167/246	medium · avg 46%	▇▇▇▇▇ 11%	sharp · at its level
Decision Index 0.4.1: BBH (accuracy) submitted	88.2% ✎	▄▆▄▄█▇▇▄▃▁▁▁ beats 227/246	easy · avg 66%	▇▇ 5.9%	typical
Decision Index 0.4.1: MMLU (accuracy) submitted	82.3% ✎	▁▄▄▃▃▂▅▇▅█▇▁ beats 163/246	easy · avg 70%	▇▇ 5.8%	sharp
Decision Index 0.4.1: GPQA Diamond (accuracy) submitted	49.5% ✎	▂▆▅▅▆█▆▃▁··▁ beats 192/246	medium · avg 39%	▇▇ 4.2%	blunt · at its level
Decision Index 0.4.1: GSM8K (accuracy (4- and 10-choice)) submitted	72.0% ✎	▁··▄█▆▆▆▅▅▃▁ beats 194/246	medium · avg 45%	▇ 3.7%	sharp · at its level
Decision Index 0.4.1: WinoGrande (accuracy) submitted	95.2% ✎	█▅█▅▄█▆▄▃▃▁▁ beats 234/246	easy · avg 78%	▇ 2.7%	typical
Decision Index 0.4.1: SATA-Bench (case exact accuracy) submitted	42.9% ✎	▁▁▁▁▁▁▁▂▅█▃▁ beats 233/246	hard · avg 22%	▇ 2.2%	at its level
Decision Index 0.4.1: ARC-Challenge (accuracy) submitted	96.2% ✎	▂▂▂▂▂▂▁▃▄▆█▆ beats 167/246	very easy · avg 90%	▇ 1.8%	sharp · maxed out

6 more tests

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: MuSR (accuracy) submitted	88.3% ✎	▁▁▂█▅▄▁▁▁▁▃▁ beats 242/246	medium · avg 60%	▇ 1.4%	blunt · at its level
Decision Index 0.4.1: CRUXEval (accuracy) submitted	81.8% ✎	██▄▃▃▄▁▁···▁ beats 202/246	medium · avg 58%	▇ 1.2%	at its level · noisy
Decision Index 0.4.1: ChessBench (accuracy) submitted	24.7% ✎	▄▇▄▄▄▆▄█▄▁▁▁ beats 219/246	hard · avg 16%	▇ 1.0%	blunt
Decision Index 0.4.1: CLadder (accuracy) submitted	93.3% ✎	▁███▆▅▃▃▃▁▁▁ beats 232/246	easy · avg 71%	▇ 1.0%	typical
Decision Index 0.4.1: ARC-Easy (accuracy) submitted	98.5% ✎	▁▁▂▁▂▂▂▂▃▆█▅ beats 157/246	very easy · avg 96%	▇ 0.7%	sharp · maxed out · noisy
Decision Index 0.4.1: HLE (accuracy) submitted	10.4% ✎	▁▁▂▃▆█▇▅▅▃▁▁ beats 41/246	very hard · avg 13%	▇ 0.0%	blunt

other tests2 not in a task · 11% of its score

test	result	vs the others in its run	difficulty	counts	why
Decision Index 0.4.1: private same skills (equated points)	59.6700 points ✎	▂▂▂▃▃▃▄▅▅▇█▃ beats 207/247	avg 41.206 points	▇▇▇ 6.9%	blunt
Decision Index 0.4.1: private new domains (equated points)	60.2300 points ✎	▃█▃▃▂▆▃▃▄▄▄▄ beats 186/247	avg 38.861 points	▇▇ 3.8%	blunt

$ where to run it

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 local run command and hardware requirements have not been verified for this entry. Source · checked 2026-10-31.

Model repository and documentation: huggingface.co/sthanika-ai ↗

$ how the score was worked out · Decision Index 0.4.1 99%

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	▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 99%	CC BY 4.0
prior · pull toward the middle		▇ 1%
weighted average		= 66.9

not counted (2)

Decision Index 0.4.1	Decision Index 0.4.1: RouterBench (selected quality (quality objective)) [correct]	79.8%	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]	46.5%	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