whatmodel.sh · find the best decision models you can run

Best Decision models for knowledge & reasoning for a 32GB graphics card

A leaderboard of the best Decision models for knowledge & reasoning for a 32GB graphics card (with 32GB of system RAM). The best one that runs on it right now is Clef-flash, followed by vLLM-SR d3-flash (9B) and Jebadiah 27B v2.1 (Ollaya Q4_K_M). Published typed-choice knowledge and reasoning evaluations. Each task score uses only the reviewed benchmarks and their original evaluator protocols. Models that run on it show the download that fits and the command to run it.

43 models · 37 benchmarks · updated

Compare tested decision models and inference recipes across text, vision and workflow tasks. Use task filters and raw results: probability consistency measures coherence, not correctness.

Filter by your hardware and budget → ~32GB fast · reset
ram[2][4][8][16][24][32][48][64][96][128][192][256]
[]
GB
gpu[apple silicon][graphics card][unified memory][none]
vram[4][8][12][16][24][32][48][64][80][96][128]
[]
GB
price[local][$0.05][$0.1][any]max per M input tokens
Filter by task
task[all tasks][tools & automation][retrieval & classification][language understanding][arts & human taste][knowledge & reasoning][vision][workflows][consistency]

Scores use reviewed knowledge & reasoning evaluations only. Compare scores within this task. Models without task measurements are excluded.

New #1s for your hardware:
orRSS

detecting your OS…

$ terminal

detecting your OS…

Run curl "whatmodel.sh/decision?ram=32&gpu=card&vram=32&max=any"to get this list in your terminal (add &format=json for agents).

Score vs release date

Release dates available for 18 of 43 models. Undated models are omitted; month-only dates are plotted at the start of the month.

#modeltask score↓where↓RAM↓speed↓get itsources
#modeltask score↓where↓RAM↓speed↓get itsources
1Jev 1.13 TypeSafe73±1api $0.042/M input tokensdocs.typesafe.ai ↗Decision Index 0.4.1LeanMCP · JevBenchDecision Models as Judges for Hallucination DetectionDecisionPivot · StartLux LabsFreyer et al. · MedQADo We Need an LLM to Know Why Sources Conflict? Decision Models for Knowledge-Conflict Typing in RAG+1
2Perplexity Decider v1.1 (27B) Perplexity73±2api price unknownopenrouter.ai ↗LeanMCP · JevBenchDecision Models as Judges for Hallucination DetectionDecisionPivot · StartLux LabsFreyer et al. · MedQA
8d1-ultra Liquid AI70±4api price unknownopenrouter.ai ↗Commit or Abstain? System One Decision Models, LLMs, and Missing Legal AuthorityDo We Need an LLM to Know Why Sources Conflict? Decision Models for Knowledge-Conflict Typing in RAGEtsabary · decision benchmarks
12d1 (hosted) Liquid AI68±2api price unknownopenrouter.ai ↗LeanMCP · JevBenchDecision Models as Judges for Hallucination DetectionFreyer et al. · MedQADo We Need an LLM to Know Why Sources Conflict? Decision Models for Knowledge-Conflict Typing in RAGEtsabary · decision benchmarks
27Solar Decide Upstage67±2api price unknownopenrouter.ai ↗LeanMCP · JevBenchFreyer et al. · MedQA
28Clef-omni Cloudflare67±1api $0.15/M input tokens~87GBdevelopers.cloudflare.com ↗Decision Index 0.4.1LeanMCP · JevBenchFreyer et al. · MedQAEtsabary · decision benchmarks
30Clef Cloudflare66±1api $0.24/M input tokens~68GBdevelopers.cloudflare.com ↗Decision Index 0.4.1LeanMCP · JevBenchDecision Models as Judges for Hallucination DetectionDecisionPivot · StartLux LabsFreyer et al. · MedQAEtsabary · decision benchmarks
35GPT-6 Luna Decisions OpenAI66±2api price unknownopenrouter.ai ↗LeanMCP · JevBenchDecision Models as Judges for Hallucination DetectionFreyer et al. · MedQACommit or Abstain? System One Decision Models, LLMs, and Missing Legal AuthorityDo We Need an LLM to Know Why Sources Conflict? Decision Models for Knowledge-Conflict Typing in RAGEtsabary · decision benchmarks
36Mercury Decide (paid endpoint) Inception66±2api price unknownopenrouter.ai ↗LeanMCP · JevBenchFreyer et al. · MedQAEtsabary · decision benchmarks
39Span-02 (Gemma 4 26B-A4B) Respan65±1api price unknownopenrouter.ai ↗Decision Index 0.4.1Etsabary · decision benchmarks
43d1.1 (hosted) Liquid AI65±7*api price unknownopenrouter.ai ↗Etsabary · decision benchmarks
47Clef-flash · bf16 Cloudflare65±1local~25GBhow to run › Decision Index 0.4.1LeanMCP · JevBenchDecisionPivot · StartLux LabsOllaya · original runsFreyer et al. · MedQAEtsabary · decision benchmarks
55Jev 1.14 TypeSafe64±4api $0.042/M input tokensdocs.typesafe.ai ↗Commit or Abstain? System One Decision Models, LLMs, and Missing Legal AuthorityDo We Need an LLM to Know Why Sources Conflict? Decision Models for Knowledge-Conflict Typing in RAGEtsabary · decision benchmarks
58Microsoft-Decision-1 Microsoft64±2api price unknownopenrouter.ai ↗LeanMCP · JevBenchFreyer et al. · MedQADo We Need an LLM to Know Why Sources Conflict? Decision Models for Knowledge-Conflict Typing in RAGEtsabary · decision benchmarks
86vLLM-SR d3-flash (9B) · ollaya-f32 vLLM Semantic Router60±1local~18GBhow to run › Decision Index 0.4.1Ollaya · original runs
94Jebadiah 27B v2.1 (Ollaya Q4_K_M) · q4_k_m-ollaya frontier-infra59±4local~16GBhow to run › Ollaya · original runs
99Perplexity Decider Flash v1.2 (4B) · ollaya-f32 Perplexity58±1local~10GBhow to run › Decision Index 0.4.1Ollaya · original runsDo We Need an LLM to Know Why Sources Conflict? Decision Models for Knowledge-Conflict Typing in RAGEtsabary · decision benchmarks
104vLLM-SR d3-mini (4B) · ollaya-f32 vLLM Semantic Router57±1local~11GBhow to run › Decision Index 0.4.1Ollaya · original runs
105Jebadiah 9B v2.1 (T1e) · bf16-t1e frontier-infra57±1†local~25GBhow to run › Decision Index 0.4.1
107Bespoke Nimble 9B v3 · ollaya-f32 Bespoke Labs57±1local~21GBhow to run › Decision Index 0.4.1Ollaya · original runs
125vLLM-SR Decision 2.0 Lux 9B · ollaya-f32 vLLM Semantic Router54±1local~20GBhow to run › Decision Index 0.4.1Ollaya · original runs
129Bespoke Nimble 9B v2 · ollaya-f32 Bespoke Labs54±1local~20GBhow to run › Decision Index 0.4.1Ollaya · original runs
137Jebadiah 9B v2.1 (Ollaya Q8_0) · q8_0-ollaya frontier-infra53±4local~9.1GBhow to run › Ollaya · original runs
139Solar Decide Flash Upstage53±2api price unknownopenrouter.ai ↗LeanMCP · JevBenchFreyer et al. · MedQAEtsabary · decision benchmarks
142Decider 4B · ollaya-f32 Mapika52±1local~11GBhow to run › Decision Index 0.4.1Ollaya · original runs
152Winnow-12B (DecisionBench b6ac22b0) · ollaya-q8-0 EldanRing51±4local~16GBhow to run › Ollaya · original runs
155vLLM-SR Decision 2.0 Nox 4B · ollaya-f32 vLLM Semantic Router51±1local~12GBhow to run › Decision Index 0.4.1Ollaya · original runs
159Kev 4B r10 · ollaya-f32 Jared Palmer50±1local~11GBhow to run › Decision Index 0.4.1Ollaya · original runs
164Jebadiah 4B v2 (Ollaya Q8_0) · q8_0-ollaya frontier-infra49±4local~5.5GBhow to run › Ollaya · original runs
172vLLM-SR d3-nano (2B) · ollaya-f32 vLLM Semantic Router47±1local~8.2GBhow to run › Decision Index 0.4.1Ollaya · original runs
187baselabs joey-1 (2B) baselabs42±1api price unknownopenrouter.ai ↗Decision Index 0.4.1Etsabary · decision benchmarks
193laya (DecisionBench 1c5edc17) · ollaya-f16 ConvAI Innovations40±4local~1.7GBhow to run › Ollaya · original runs
197vLLM-SR d3-lite (0.8B) · ollaya-f32 vLLM Semantic Router40±1local~4.5GBhow to run › Decision Index 0.4.1Ollaya · original runs
201vLLM-SR Decision 2.0 Sol 2B · ollaya-f32 vLLM Semantic Router37±1local~11GBhow to run › Decision Index 0.4.1Ollaya · original runs
203Strands Decider 2B (Hobson v21) · bf16 Amazon Web Services37±1local~6.6GBhow to run › Decision Index 0.4.1
204GLiClass v4 edge (0.4B) · ollaya-f32 Knowledgator Engineering37±1local~2.7GBhow to run › Decision Index 0.4.1Ollaya · original runs
217Kev 0.8B r15 · ollaya-f32 Jared Palmer33±1local~4.6GBhow to run › Decision Index 0.4.1Ollaya · original runs
223vLLM-SR d3-edge (0.6B) · ollaya-f32 vLLM Semantic Router30±1local~2.6GBhow to run › Decision Index 0.4.1Ollaya · original runs
225vLLM-SR Decision 2.0 Eos 0.8B · ollaya-f32 vLLM Semantic Router30±1local~5.8GBhow to run › Decision Index 0.4.1Ollaya · original runs
228vLLM-SR Decision 2.0 Kai 0.6B · ollaya-f32 vLLM Semantic Router29±1local~3.7GBhow to run › Decision Index 0.4.1Ollaya · original runs
231LiquidAI d1-omni-600M · ollaya-f32 Liquid AI28±1local~2.7GBhow to run › Decision Index 0.4.1Ollaya · original runs
237Decision-1.0-Eos-0.8B (DecisionBench 3c2d6326) · ollaya-f32 llm-semantic-router26±3local~5.8GBhow to run › Ollaya · original runs
245CLM-v0.1-8B Qwen3-8B heads · ollaya-f32 Contrastive-LM23±1local~15GBhow to run › Decision Index 0.4.1Ollaya · original runs

+ 220 more don't fit your machine or budget · show them

263 of 263 catalogue models listed here · sources and licences

* new or thinly tested: ranked on too few benchmark results · † no independently run results verified · ~RAM is estimated; full context may need more memory

43 models · 37 benchmarks · updated