ProofBench
Das Ergebnis jedes Modells, so, wie der Benchmarking Hub von Epoch AI es veröffentlicht, von Epoch gemessen oder aus der genannten Quelle gesammelt. Facthory hat sie nicht selbst gemessen.
Daten: Epoch AI, „Capabilities & benchmarking“, epoch.ai/benchmarks, lizenziert unter CC BY 4.0. Gelesen am 16.11.2026.
Das Diagramm zeigt die 25 besten von 85 Ergebnissen; die Tabelle führt alle auf.
| # | Modell | Organisation | Ergebnis | Nachweis |
|---|---|---|---|---|
| 1 | Claude Fable 5.1 (max) API access | Anthropic | 100 % | – |
| 2 | Claude Fable 5.2 (max) API access | Anthropic | 100 % | – |
| 3 | Claude Opus 5.5 (max) API access | Anthropic | 100 % | – |
| 4 | Claude Sonnet 5.5 (max) API access | Anthropic | 100 % | – |
| 5 | Claude Opus 5 (max) API access | Anthropic | 99 % | – |
| 6 | Gemini 4 Argon (unknown) API access | Google DeepMind | 99 % | – |
| 7 | GPT-6 Astra (unknown thinking) API access | OpenAI | 99 % | – |
| 8 | GPT-6.1 Sol (unknown thinking) API access | OpenAI | 99 % | – |
| 9 | Claude Fable 5 (max) API access | Anthropic | 95 % | – |
| 10 | Kimi K3.1 Pro (unknown) Open weights (restricted use) | Moonshot | 91 % | – |
| 11 | Claude Haiku 5.5 (max) API access | Anthropic | 89 % | – |
| 12 | Kimi K3 (unknown) Open weights (non-commercial) | Moonshot | 87 % | – |
| 13 | Muse Wonder (unknown) API access | Meta AI | 84 % | – |
| 14 | GPT-5.6 Sol (max) API access | OpenAI | 83 % | – |
| 15 | GPT-6 Sol (unknown thinking) API access | OpenAI | 83 % | – |
| 16 | Claude Sonnet 5 (max) API access | Anthropic | 77 % | – |
| 17 | Tencent Hy4 preview (unknown) Open weights (unrestricted) | Tencent | 75 % | – |
| 18 | GPT-5.6 Terra (xhigh) API access | OpenAI | 74 % | – |
| 19 | MiMo-V2.6 Pro | Xiaomi Corp | 70 % | – |
| 20 | Claude Opus 4.8 API access | Anthropic | 69 % | – |
| 21 | GPT-6 Luna (unknown thinking) API access | OpenAI | 64 % | – |
| 22 | MiMo-V2.6 Flash | Xiaomi Corp | 63 % | – |
| 23 | GPT-5.6 Luna (max) API access | OpenAI | 60 % | – |
| 24 | Gemini 3.7 Flash (unknown) API access | Google DeepMind | 58 % | – |
| 25 | Muse Spark 1.3 (max) API access | Meta AI | 58 % | – |
| 26 | Qwen3.8 Max (unknown) API access | Alibaba | 58 % | – |
| 27 | DeepSeek V4 Flash 0731 (unknown) Open weights (unrestricted) | DeepSeek | 56 % | – |
| 28 | GPT-5.4 (xhigh) API access | OpenAI | 56 % | – |
| 29 | Muse Spark 1.3 (unknown) API access | Meta AI | 55 % | – |
| 30 | Claude Opus 4.7 (max) API access | Anthropic | 54 % | – |
| 31 | DeepSeek V4.1 Flash (unknown) Open weights (unrestricted) | DeepSeek | 54 % | – |
| 32 | Grok 4.6 (unknown) API access | xAI | 51 % | – |
| 33 | Claude Opus 4.6 (max) API access | Anthropic | 50 % | – |
| 34 | DeepSeek V4 Pro 0813 (unknown) Open weights (unrestricted) | DeepSeek | 50 % | – |
| 35 | GPT-5.5 (xhigh) API access | OpenAI | 50 % | – |
| 36 | GLM-5.3 (max) Open weights (restricted use) | Z.ai (Zhipu AI) | 49 % | – |
| 37 | Gemini 3.8 Flash (unknown) API access | Google DeepMind | 48 % | – |
| 38 | Claude Sonnet 4.6 (max) API access | Anthropic | 45 % | – |
| 39 | Muse Spark 1.2 API access | Meta AI | 43 % | – |
| 40 | Step 5 Preview (unknown) API access | StepFun | 42 % | – |
| 41 | Muse Spark 1.1 API access | Meta AI | 39 % | – |
| 42 | Claude Opus 4.5 (unknown thinking) API access | Anthropic | 36 % | – |
| 43 | Gemini 3.6 Flash API access | Google DeepMind | 36 % | – |
| 44 | GLM-5.2 Open weights (unrestricted) | Z.ai (Zhipu AI) | 35 % | – |
| 45 | Gemini 3.5 Flash API access | Google DeepMind | 31 % | – |
| 46 | Grok 4.5 (high) API access | xAI | 31 % | – |
| 47 | Gemini 3.1 Pro Preview API access | Google DeepMind | 26 % | – |
| 48 | Grok 4.7 (unknown) API access | xAI | 26 % | – |
| 49 | Qwen3.7 Max API access | Alibaba | 26 % | – |
| 50 | GLM-5.1 Open weights (unrestricted) | Z.ai (Zhipu AI) | 22,2 % | – |
| 51 | mimo-v2.5-pro Open weights (unrestricted) | Xiaomi Corp | 22 % | – |
| 52 | GLM-5.3-Flash (max) Open weights (unrestricted) | Z.ai (Zhipu AI) | 21 % | – |
| 53 | GPT-5.4 mini (xhigh) API access | OpenAI | 21 % | – |
| 54 | Gemini 3 Pro Preview API access | Google DeepMind | 20 % | – |
| 55 | Claude Sonnet 4.5 (unknown thinking) API access | Anthropic | 19 % | – |
| 56 | MiniMax-M3 Open weights (restricted use) | MiniMax | 18 % | – |
| 57 | Beam 1.0 (unknown) Open weights (unrestricted) | Reflection | 18 % | – |
| 58 | GPT-5 (high) API access | OpenAI | 18 % | – |
| 59 | Muse Spark API access | Meta AI | 17 % | – |
| 60 | DeepSeek v4 Pro (max) Open weights (unrestricted) | DeepSeek | 16 % | – |
| 61 | Kimi K2.6 Open weights (unrestricted) | Moonshot | 16 % | – |
| 62 | mimo-v2.5 Open weights (unrestricted) | Xiaomi Corp | 16 % | – |
| 63 | Qwen3.8 27B Open weights (unrestricted) | Alibaba | 16 % | – |
| 64 | Gemini 3 Flash Preview API access | Google DeepMind | 15 % | – |
| 65 | GPT-5.2 (xhigh) API access | OpenAI | 15 % | – |
| 66 | grok-4.20-0309-reasoning API access | xAI | 14 % | – |
| 67 | Gemini 3.5 Flash-Lite API access | Google DeepMind | 13 % | – |
| 68 | GPT-5 nano (high) API access | OpenAI | 12 % | – |
| 69 | grok-4.3_high Hosted access (no API) | xAI | 11 % | – |
| 70 | Mistral Large 4 Preview (unknown) Open weights (restricted use) | Mistral AI | 10 % | – |
| 71 | GPT-5 mini (high) API access | OpenAI | 9 % | – |
| 72 | GPT-5.1-Codex-Max API access | OpenAI | 9 % | – |
| 73 | Mistral Medium 3.5 Open weights (restricted use) | Mistral AI | 9 % | – |
| 74 | DeepSeek-V3.2 (Thinking; Fireworks) Open weights (unrestricted) | DeepSeek | 8 % | – |
| 75 | GLM-4.7 Open weights (unrestricted) | Z.ai (Zhipu AI) | 6 % | – |
| 76 | Inkling Small Open weights (unrestricted) | Thinking Machines | 6 % | – |
| 77 | GPT-5.4 nano (high) API access | OpenAI | 5 % | – |
| 78 | MiniMax-M2.5 Open weights (unrestricted) | MiniMax | 4 % | – |
| 79 | grok-4-1-fast-reasoning API access | xAI | 4 % | – |
| 80 | MiniMax-M2.7 Open weights (non-commercial) | MiniMax | 3 % | – |
| 81 | Mercury 2.5 (unknown) API access | Inception Labs | 3 % | – |
| 82 | nemotron-3-ultra Open weights (unrestricted) | Nvidia | 2 % | – |
| 83 | Inkling Open weights (unrestricted) | Thinking Machines | 0 % | – |
| 84 | Laguna M.1 | Poolside | 0 % | – |
| 85 | Laguna XS.2 Open weights (unrestricted) | Poolside | 0 % | – |
