whatmodel.sh / decision / Strands Decider 2B (Hobson v21)

Strands Decider 2B (Hobson v21)

+ contribute
Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

Strands Decider 2B (Hobson v21) is an open-weights decision model from Amazon Web Services, released October 2026. Its published size is 1.9B parameters. Licence: apache-2.0.

Catalogue identity notes

Exact Decision Index 0.4.1 text-board engine strands-decider-2b. Produces probabilities over supplied choices, not conversational text. Only public per-benchmark results are used; no Full index, private scores or latency is treated as an ability measurement. Model and runtime documentation reviewed 2026-10-31; no fresh hardware execution. LoRA adapter plus decision head on Qwen3.5-2B-Base; size is the approximately 1.9B served model, not the adapter download alone.

Model information sources

Release date ↗ · Weights licence ↗ · Parameter count ↗

HF safetensors metadata for full weights; Strands uses its documented approximate served backbone size. Memory estimate: BF16 weights plus 20% runtime allowance and 2 GB; not a measured requirement.

35.6±1.1#281 of 355 decision modelsfrom 42 benchmark results across 1 source

$ how it compares

The decision models ranked either side of Strands Decider 2B (Hobson v21), out of 357 with a score.

#modelscorewhereRAMreleased
281Sieve 2B Sthānika AI37±1model files2026-10-13
282smalljev-semantic-v9 (DecisionBench 8e2bac7c) isHeSatoshi36±2model files
283Strands Decider 2B (Hobson v21) Amazon Web Services36±1local~6.6GB2026-10
284vLLM-SR vela3-fast-vision (Vela 3.0, 0.25B) vLLM Semantic Router35±1model files
285wyat-1-800m teilomillet34±1model files2026-10-18

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

$ what it's good at

tools & automation▁▃▁▁▁▂▃▄▆██▆ #202 of 250 · 9.4% of its score · 6 tests

Beats Cam1, mini-jev Qwen3-4B (decision readout), InternLM Intern-Decision Qwen3.5-2B here, though they rank higher overall. Trails JevK5 Lite, Mola-68M, Crivo-68M here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
Decision Index 0.4.1: BFCL (case exact accuracy)81.3% ✎▁▂▁▁▁▃▂▄▅▆█▁ beats 47/246very easy · avg 92%▇▇ 5.0%sharp · noisy
Decision Index 0.4.1: API-Bank (accuracy)39.6% ✎▁▁▁▁▁▁▁▃▃▂█▂ beats 29/236easy · avg 70%▇ 2.3%sharp · at its level · noisy
Decision Index 0.4.1: When2Call (accuracy)59.6% ✎▁▁▁▂▃▅▅▆█▇█▃ beats 65/246easy · avg 72%▇ 1.1%at its level · noisy
Decision Index 0.4.1: ToolRet (nDCG@10)60.5% ✎▁▁▁▁▁·▁▁▁▄█▁ beats 69/246medium · avg 60%▇ 0.8%blunt · at its level
Decision Index 0.4.1: Home appliance simulator (case exact accuracy)0.0% ✎█▁▁▃▄▄▆▄▁▂▁▃ beats 49/243medium · avg 37%▇ 0.3%sharp · noisy
Decision Index 0.4.1: RouterBench (selected quality (quality objective))78.9% ✎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▁▁▁·▁▃▁▁▃▃█▅ #235 of 258 · 4.8% of its score · 7 tests

Beats Sieve 2B here, though they rank higher overall. Trails MoganAI Lavoir ModernBERT-large, Bekko System One v0 68M, Bosun v3.1 0.6B LoRA here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
Decision Index 0.4.1: Amazon ESCI (macro-F1)26.0% ✎▁▁▁▁▁▁▃▂▃▃█▅ beats 21/246medium · avg 47%▇ 1.5%at its level
Decision Index 0.4.1: BRIGHT (nDCG@10)36.8% ✎▁▁▁▁▁▁▁▂▂▄▇█ beats 80/246medium · avg 38%▇ 1.4%blunt · at its level
Decision Index 0.4.1: CLINC150+OOS (macro-F1)64.7% ✎▁▁▁▁▁▁▁▃▃▆█▄ beats 36/238easy · avg 83%▇ 0.6%sharp · at its level · noisy
Decision Index 0.4.1: HoVer (accuracy)54.0% ✎▆▆▇▅▄▄▃▇▄█▆▄ beats 21/246easy · avg 74%▇ 0.5%blunt · at its level
Decision Index 0.4.1: PhishNChips phishing decisions (accuracy)50.8% ✎█▅▅▄▃▂▁▁▁▁·▁ beats 44/245easy · avg 70%▇ 0.5%at its level · noisy
Decision Index 0.4.1: BANKING77 (macro-F1)80.4% ✎▁·▁▁▁▁▁▁▂█▇█ beats 101/238easy · avg 79%▇ 0.3%at its level · noisy
Decision Index 0.4.1: SGD/SGD-X (macro-F1)59.6% ✎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▂▂▂▃▂▅▆▃█▆▄▃ #216 of 256 · 24% of its score · 9 tests

Beats Saina Helm 2 Omni (0.8B), Gevva E2B here, though they rank higher overall. Trails LiquidAI d1-omni-600M, DecisionTune 1.0, vera-spark here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
Decision Index 0.4.1: HellaSwag (accuracy)57.4% ✎▂▄▃▂▄▇▃▇█▅▃▁ beats 40/246very easy · avg 85%▇▇▇▇ 9.4%sharp · at its level
Decision Index 0.4.1: NLI4CT (macro-F1)65.8% ✎▁▁▁▂▂▃▃▄▇▇█▄ beats 61/246easy · avg 74%▇▇ 5.3%at its level
Decision Index 0.4.1: ANLI (macro-F1)39.9% ✎▁▂▃▅▅█▂····▁ beats 35/246medium · avg 60%▇▇ 4.1%at its level
Decision Index 0.4.1: iSarcasmEval (Sarcasm F1 · track A, English)29.6% ✎▁·▁·▁▁▂▄▅█▇▄ beats 58/246medium · avg 44%▇ 1.3%at its level
Decision Index 0.4.1: ContractNLI (macro-F1)73.0% ✎▁▁▁▁▁▁▁▂▂▇▇█ beats 96/246easy · avg 71%▇ 1.0%at its level · noisy
Decision Index 0.4.1: FinEntity (macro-F1)72.7% ✎▁···▁▁▁▁▃▆█▃ beats 50/246easy · avg 84%▇ 0.9%at its level · noisy
Decision Index 0.4.1: VAST (macro-F1)40.4% ✎▁▁▁▂▅▄▄▂▄▆█▄ beats 31/246medium · avg 62%▇ 0.8%at its level
Decision Index 0.4.1: RAGTruth (F1 on hallucinated class)35.7% ✎▁▁▁▁▁▁▁▃▁▃▆█ beats 35/246easy · avg 66%▇ 0.7%at its level · noisy
1 more test
testresultvs the others in its rundifficultycountswhy
Decision Index 0.4.1: ACOS (per-review F1)6.0% ✎▁··▂▃▃▅▇█▇▃▃ beats 31/246hard · avg 22%▇ 0.4%noisy
arts & human taste▁▇▄▃▅▅█▅▄▂▁▁ #205 of 247 · 7.5% of its score · 6 tests

Beats Valen-4B, Saracura PT-BR 4B, TokenRhythm NeoHorse-Jev-4B here, though they rank higher overall. Trails LFM2.5-350M-RLCD, system-one-gemma, Qwen-2.5-1B-RLCD (decision readout) here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
Decision Index 0.4.1: New Yorker (accuracy)45.5% ✎▁▁▂▃▃▃▇▅█▅▃▁ beats 51/246medium · avg 61%▇ 3.2%at its level
Decision Index 0.4.1: BPoMP (accuracy)56.0% ✎▁█▄▄▅█▆▅▄▁▁▁ beats 42/246easy · avg 83%▇ 1.7%sharp · at its level · noisy
Decision Index 0.4.1: Humicroedit (accuracy)49.7% ✎▁·▁▃▂▂▅█▄▂▁▁ beats 19/246medium · avg 60%▇ 1.4%blunt · at its level
Decision Index 0.4.1: cfcolor (accuracy)50.7% ✎▁········▁▇█ beats 23/246medium · avg 60%▇ 0.8%blunt · at its level
Decision Index 0.4.1: Habermas Machine (accuracy)38.0% ✎▁▂▄█▇▅▁▁▄▄▁▁ beats 40/246medium · avg 47%▇ 0.2%blunt · at its level
Decision Index 0.4.1: POP909-CL (accuracy)6.9% ✎▄▅▅▃▆▅▃▅▂▅▃█ beats 85/237hard · avg 21%▇ 0.2%sharp · too hard for it · noisy
knowledge & reasoning▁▂▂▂▃▃▃▅▅▆█▂ #203 of 263 · 48% of its score · 14 tests

Beats GLiClass v4 edge (0.4B), Crivo-1B, vLLM-SR vela3-fast (Vela 3.0, 0.4B) here, though they rank higher overall. Trails laya (DecisionBench 1c5edc17), DM-JEPA 1.1 here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
Decision Index 0.4.1: ARC-Challenge (accuracy)71.1% ✎▂▂▂▂▃▂▁▃▄▆█▆ beats 55/246very easy · avg 90%▇▇▇▇ 10%sharp · at its level
Decision Index 0.4.1: ARC-Easy (accuracy)90.5% ✎▁▁▂▁▂▂▃▂▃▆█▅ beats 67/246very easy · avg 96%▇▇▇▇ 9.0%sharp · noisy
Decision Index 0.4.1: MMLU (accuracy)51.0% ✎▁▄▄▃▃▂▅▇▅█▇▁ beats 60/246easy · avg 70%▇▇▇ 6.9%sharp · at its level
Decision Index 0.4.1: MMLU-Pro (accuracy)31.1% ✎▄▄▃▃▃▄▅▇██▅▁ beats 66/246medium · avg 46%▇▇ 5.8%sharp · at its level
Decision Index 0.4.1: BBH (accuracy)46.5% ✎▄▆▄▄█▇▇▄▃▁▁▁ beats 56/246easy · avg 66%▇▇ 5.6%at its level
Decision Index 0.4.1: WinoGrande (accuracy)57.4% ✎█▅█▅▄█▆▄▃▃▁▁ beats 37/246easy · avg 78%▇ 3.5%at its level
Decision Index 0.4.1: GPQA Diamond (accuracy)33.2% ✎▂▆▅▅▆█▆▃▁··▁ beats 77/246medium · avg 39%▇ 1.9%blunt · at its level
Decision Index 0.4.1: GSM8K (accuracy (4- and 10-choice))24.7% ✎▁··▄█▆▆▆▅▅▃▁ beats 55/246medium · avg 45%▇ 1.8%sharp
6 more tests
testresultvs the others in its rundifficultycountswhy
Decision Index 0.4.1: MuSR (accuracy)51.1% ✎▁▁▃█▅▄▁▁▁▁▁▁ beats 39/246medium · avg 60%▇ 0.9%blunt · at its level
Decision Index 0.4.1: CLadder (accuracy)54.4% ✎▁███▆▅▃▃▁▁▁▁ beats 36/246easy · avg 71%▇ 0.9%at its level
Decision Index 0.4.1: CRUXEval (accuracy)33.3% ✎██▄▃▃▄▁▁···▁ beats 15/246medium · avg 58%▇ 0.8%at its level · noisy
Decision Index 0.4.1: SATA-Bench (case exact accuracy)9.8% ✎▁▁▁▁▁▁▃▂▅█▂▁ beats 43/246hard · avg 22%▇ 0.6%typical
Decision Index 0.4.1: ChessBench (accuracy)9.2% ✎▄▇▄▄▄▆▄█▄▁▁▁ beats 48/246hard · avg 16%▇ 0.3%blunt
Decision Index 0.4.1: HLE (accuracy)14.0% ✎▁▁▂▃▆█▇▅▅▃▁▁ beats 176/246very hard · avg 13%▇ 0.0%blunt
other tests2 not in a task · 5.9% of its score
testresultvs the others in its rundifficultycountswhy
Decision Index 0.4.1: private same skills (equated points)21.6400 points ✎▂▂▂▃▃▃▄▅▅▇█▃ beats 42/247avg 41.206 points▇▇ 3.8%blunt
Decision Index 0.4.1: private new domains (equated points)19.3400 points ✎▃█▃▃▂▆▃▃▄▄▄▄ beats 72/247avg 38.861 points▇ 2.1%blunt

$ where to run it

Other ways to run it

  1. Run it as published · needs ~6.6 GB, NVIDIA only · from huggingface.co/StrandsAgents
    pip install strands-decider && strands-decider serve StrandsAgents/strands-decider-2B-hobson-v21 --port 8000

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. See the local setup below for the reviewed runtime. Source · checked 2026-10-31.

Model repository and documentation: huggingface.co/StrandsAgents ↗

$ how the score was worked out · Decision Index 0.4.1 100%

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

sourceresultsshare of the scorelicence
Decision Index 0.4.142CC BY 4.0
prior · pull toward the middle
weighted average= 35.6

not counted (2)

Decision Index 0.4.1Decision Index 0.4.1: RouterBench (selected quality (quality objective)) [correct]78.9%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 0.4.1Decision Index 0.4.1: SGD/SGD-X (macro-F1) [correct]59.6%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 informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit 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 · sources and licences · json