Reasoning
SimpleBench Leaderboard
SimpleBench is a set of everyday, common-sense and trick questions that humans answer easily but language models often get wrong. It probes basic reasoning and robustness rather than specialist knowledge.
Source: epoch25 open models ranked+76 proprietaryData through Aug 2026
All models ranked on SimpleBench
Proprietary / closed models are shown dimmed — you can't run them locally, but they show where the open field stands.
| # | Model | Score | VRAM @ Q4 |
|---|---|---|---|
| 1 | Claude Fable 5 Max · proprietary | 81.9% | API only |
| 2 | Claude Opus 5 (unspecified) · proprietary | 80.6% | API only |
| 3 | Gemini 3.1 Pro Preview · proprietary | 79.6% | API only |
| 4 | GPT 5.5 Pro (high) · proprietary | 76.9% | API only |
| 5 | GPT 5.5 Pro (unspecified) · proprietary | 76.9% | API only |
| 6 | Gemini 3.5 Flash (unspecified) · proprietary | 76.7% | API only |
| 7 | Gemini 3 Pro Preview · proprietary | 76.4% | API only |
| 8 | Grok 4.6 (unspecified) · proprietary | 75.9% | API only |
| 9 | Muse Spark 1.2 · proprietary | 74.5% | API only |
| 10 | GPT 5.4 Pro (Mar 05, 2026, unspecified) · proprietary | 74.1% | API only |
| 11 | GPT 5.6 Sol Prounknown · proprietary | 71.7% | API only |
| 12 | GPT 5.6 Sol Proxhigh · proprietary | 71.7% | API only |
| 13 | Qwen3.7 Max · proprietary | 70.4% | API only |
| 14 | Grok 4.5 (unspecified) · proprietary | 70.0% | API only |
| 15 | GPT 5.5 (unspecified) · proprietary | 69.0% | API only |
| 16 | Claude Opus 4.6 · proprietary | 67.6% | API only |
| 17 | Claude Opus 4.8 (unspecified) · proprietary | 64.8% | API only |
| 18 | GPT 5.6 Sol (unspecified) · proprietary | 64.8% | API only |
| 19 | GPT 5.6 Sol (xhigh) · proprietary | 64.8% | API only |
| 20 | Qwen3.6 Max Preview · proprietary | 63.0% | API only |
| 21 | Gemini 2.5 Pro Preview (Jun 05) · proprietary | 62.4% | API only |
| 22 | Claude Opus 4.5 (Nov 01, 2025, unspecified) · proprietary | 62.0% | API only |
| 23 | Claude Opus 4.5 (Nov 01, 2025) · proprietary | 62.0% | API only |
| 24 | Claude Opus 4.7 (unspecified) · proprietary | 61.7% | API only |
| 25 | GPT 5 Pro (Oct 06, 2025, high) · proprietary | 61.6% | API only |
| 26 | GPT 5 Pro (Oct 06, 2025, unspecified) · proprietary | 61.6% | API only |
| 27 | DeepSeek V4 Flash 0731 · 304.2B | 61.1% | ~183 GB |
| 28 | Gemini 3 Flash Preview · proprietary | 61.1% | API only |
| 29 | Kimi K3 · 2779.9B | 60.7% | ~1674 GB |
| 30 | Claude Sonnet 5 (unspecified) · proprietary | 60.6% | API only |
| 31 | Grok 4 (Jul 09) · proprietary | 60.5% | API only |
| 32 | Claude Opus 4.1 (Aug 05, 2025, unspecified) · proprietary | 60.0% | API only |
| 33 | Claude Opus 4.1 (Aug 05, 2025) · proprietary | 60.0% | API only |
| 34 | Claude Opus 4 (May 14, 2025, 12K) · proprietary | 58.8% | API only |
| 35 | Claude Opus 4 (May 14, 2025, unspecified) · proprietary | 58.8% | API only |
| 36 | GLM 5.2 · 753.3B | 58.8% | ~456 GB |
| 37 | Kimi K2.7 Code · 1026.9B | 57.9% | ~620 GB |
| 38 | GPT 5.2 Pro (Dec 11, 2025) · proprietary | 57.4% | API only |
| 39 | GPT 5 (Aug 07, 2025, high) · proprietary | 56.7% | API only |
| 40 | Grok 4.1 Fast Non Reasoning · proprietary | 56.0% | API only |
| 41 | Grok 4.1 Fast Reasoning · proprietary | 56.0% | API only |
| 42 | GLM 5.1 · 753.9B | 55.1% | ~457 GB |
| 43 | Claude Sonnet 4.5 (Sep 29, 2025, 12K) · proprietary | 54.3% | API only |
| 44 | Claude Sonnet 4.5 (Sep 29, 2025, unspecified) · proprietary | 54.3% | API only |
| 45 | GLM 5 · 753.9B | 53.2% | ~457 GB |
| 46 | GPT 5.1 (Nov 13, 2025, high) · proprietary | 53.2% | API only |
| 47 | O3 (Apr 16, 2025, high) · proprietary | 53.1% | API only |
| 48 | DeepSeek V3.2 Speciale · 685.4B | 52.6% | ~415 GB |
| 49 | Gemini 2.5 Pro Exp (Mar 25) · proprietary | 51.6% | API only |
| 50 | Gemini 2.5 Pro Preview (Mar 25) · proprietary | 51.6% | API only |
| 51 | Inkling · 952.4B | 50.0% | ~572 GB |
| 52 | GPT 5.6 Terra (unspecified) · proprietary | 48.9% | API only |
| 53 | GPT 5.6 Terra (xhigh) · proprietary | 48.9% | API only |
| 54 | GLM 4.7 · 358.3B | 47.7% | ~216 GB |
| 55 | GPT 5.6 Luna (unspecified) · proprietary | 46.8% | API only |
| 56 | GPT 5.6 Luna (xhigh) · proprietary | 46.8% | API only |
| 57 | Kimi K2.5 · 1026.9B | 46.8% | ~620 GB |
| 58 | Claude 3.7 Sonnet (Feb 19, 2025, 12K) · proprietary | 46.4% | API only |
| 59 | Claude 3.7 Sonnet (Feb 19, 2025, unspecified) · proprietary | 46.4% | API only |
| 60 | DeepSeek V4 Flash · 290.9B | 46.3% | ~175 GB |
| 61 | GPT 5.2 (Dec 11, 2025, high) · proprietary | 45.8% | API only |
| 62 | GPT 5.2 (Dec 11, 2025, unspecified) · proprietary | 45.8% | API only |
| 63 | MiniMax M3 · 427.0B | 45.8% | ~257 GB |
| 64 | Claude Sonnet 4 (May 14, 2025, 12K) · proprietary | 45.5% | API only |
| 65 | Claude Sonnet 4 (May 14, 2025, unspecified) · proprietary | 45.5% | API only |
| 66 | Claude 3.7 Sonnet (Feb 19, 2025) · proprietary | 44.9% | API only |
| 67 | O1 Preview (Sep 12, 2024) · proprietary | 41.7% | API only |
| 68 | Claude 3.5 Sonnet (Oct 22, 2024) · proprietary | 41.4% | API only |
| 69 | Gemini 2.5 Flash · proprietary | 41.2% | API only |
| 70 | DeepSeek R1 0528 · 684.5B | 40.8% | ~415 GB |
| 71 | O1 (Dec 17, 2024, high) · proprietary | 40.1% | API only |
| 72 | DeepSeek V3.1 · 684.5B | 40.0% | ~415 GB |
| 73 | O4 Mini (Apr 16, 2025, high) · proprietary | 38.7% | API only |
| 74 | O1 (Dec 17, 2024, medium) · proprietary | 36.7% | API only |
| 75 | Grok 3 · proprietary | 36.1% | API only |
| 76 | Qwen3.6 Flash · proprietary | 35.2% | API only |
| 77 | GPT 4.5 Preview (Feb 27, 2025) · proprietary | 34.5% | API only |
| 78 | Gemini Exp (Dec 06) · proprietary | 31.1% | API only |
| 79 | Qwen3 235B A22B · 235.1B | 31.0% | ~142 GB |
| 80 | DeepSeek R1 · 684.5B | 30.9% | ~415 GB |
| 81 | Gemini 2.0 Flash Thinking Exp (Jan 21) · proprietary | 30.7% | API only |
| 82 | Llama 4 Maverick 17B 128E Instruct · 401.6B | 27.7% | ~265 GB |
| 83 | Claude 3.5 Sonnet (Jun 20, 2024) · proprietary | 27.5% | API only |
| 84 | DeepSeek v3 0324 · 684.5B | 27.2% | ~415 GB |
| 85 | Gemini 1.5 Pro 002 · proprietary | 27.1% | API only |
| 86 | GPT 4.1 (Apr 14, 2025) · proprietary | 27.0% | API only |
| 87 | Kimi K2 Instruct · 1026.4B | 26.3% | ~620 GB |
| 88 | GPT 4 Turbo (Apr 09, 2024) · proprietary | 25.1% | API only |
| 89 | Claude 3 Opus (Feb 29, 2024) · proprietary | 23.5% | API only |
| 90 | Llama 3.1 405B Instruct · 405.9B | 23.0% | ~268 GB |
| 91 | O3 Mini (Jan 31, 2025, high) · proprietary | 22.8% | API only |
| 92 | Grok 2 (Dec 12) · proprietary | 22.7% | API only |
| 93 | Mistral Large Instruct 2407 · 122.6B | 22.5% | ~81 GB |
| 94 | GPT OSS 120B · 116.8B | 22.1% | ~71 GB |
| 95 | Llama 3.3 70B Instruct · 70.6B | 19.9% | ~47 GB |
| 96 | DeepSeek v3 · 684.5B | 18.9% | ~415 GB |
| 97 | Gemini 2.0 Flash Exp · proprietary | 18.9% | API only |
| 98 | O1 Mini (Sep 12, 2024, medium) · proprietary | 18.1% | API only |
| 99 | GPT 4o (Aug 06, 2024) · proprietary | 17.8% | API only |
| 100 | C4ai Command R Plus 08 2024 · 103.8B | 17.4% | ~69 GB |
| 101 | GPT 4o Mini (Jul 18, 2024) · proprietary | 10.7% | API only |
Score vs model size
Which models give the most quality for their size — the ones worth running locally.
- Llama 3.3 70B Instruct, 71B, score 19.9% — on the efficiency frontier (best score at its size or smaller).
- GPT OSS 120B, 117B, score 22.1% — on the efficiency frontier (best score at its size or smaller).
- Mistral Large Instruct 2407, 123B, score 22.5% — on the efficiency frontier (best score at its size or smaller).
- Qwen3 235B A22B, 235B, score 31.0% — on the efficiency frontier (best score at its size or smaller).
- DeepSeek V4 Flash, 291B, score 46.3% — on the efficiency frontier (best score at its size or smaller).
- DeepSeek V4 Flash 0731, 304B, score 61.1% — on the efficiency frontier (best score at its size or smaller).
SimpleBench: frequently asked questions
- What is the best open LLM on SimpleBench?
- DeepSeek V4 Flash 0731 is the top open model on SimpleBench, scoring 61.1%. Among all models tested — including proprietary ones — it ranks #27. The top model overall is Claude Fable 5 Max (Anthropic) at 81.9%.
- Can open models match proprietary models on SimpleBench?
- Not quite on SimpleBench: the strongest proprietary model (Claude Fable 5 Max) scores 81.9%, ahead of the best open model (DeepSeek V4 Flash 0731) at 61.1% — but you can run the open one yourself.
Scores aggregated from epoch. llmrun does not run this benchmark — see the source for methodology, or the about benchmarks for what it measures.