WeirdML v3
Each model’s result as Epoch AI’s Benchmarking Hub publishes it, run by Epoch or collected from the source it cites. Facthory has not run them.
Data: Epoch AI, ‘Capabilities & benchmarking’, epoch.ai/benchmarks, licensed CC BY 4.0. Read on 16 Nov 2026.
The chart shows the 25 best of 26 results; the table lists them all.
| # | Model | Organization | Result | Evidence |
|---|---|---|---|---|
| 1 | GPT-6 Astra (xhigh) API access | OpenAI | 54% | – |
| 2 | GPT-6.1 Sol (xhigh) API access | OpenAI | 50.6% | – |
| 3 | Claude Opus 5.5 (xhigh) API access | Anthropic | 49.8% | – |
| 4 | Claude Fable 5.1 (xhigh) API access | Anthropic | 43.5% | – |
| 5 | Gemini 4 Argon (high) API access | Google DeepMind | 41.4% | – |
| 6 | GPT-6.1 Terra (xhigh) API access | OpenAI | 40.4% | – |
| 7 | Claude Sonnet 5.5 (xhigh) API access | Anthropic | 37% | – |
| 8 | Claude Opus 5 (xhigh) API access | Anthropic | 32.5% | – |
| 9 | Claude Fable 5 (xhigh) API access | Anthropic | 27.9% | – |
| 10 | GPT-6 Sol (xhigh) API access | OpenAI | 27.6% | – |
| 11 | Muse Wonder (xhigh) API access | Meta AI | 24.9% | – |
| 12 | Claude Haiku 5.5 (xhigh) API access | Anthropic | 24.1% | – |
| 13 | GPT-5.6 Sol (xhigh) API access | OpenAI | 21.4% | – |
| 14 | DeepSeek V4.1 Flash 1017 (high) Open weights (unrestricted) | DeepSeek | 19.9% | – |
| 15 | Kimi K3.1 Pro (xhigh) Open weights (restricted use) | Moonshot | 18.9% | – |
| 16 | Grok 4.7 (xhigh) API access | xAI | 18.7% | – |
| 17 | Kimi K3 (high) Open weights (non-commercial) | Moonshot | 15.4% | – |
| 18 | Gemini 3.8 Flash (high) API access | Google DeepMind | 14.6% | – |
| 19 | Mistral Large 4 Preview (high) Open weights (restricted use) | Mistral AI | 12.2% | – |
| 20 | Muse Spark 1.3 (xhigh) API access | Meta AI | 12.1% | – |
| 21 | GPT-6 Luna (xhigh) API access | OpenAI | 12% | – |
| 22 | DeepSeek V4.1 Flash (high) Open weights (unrestricted) | DeepSeek | 11.6% | – |
| 23 | GPT-5.2 (xhigh) API access | OpenAI | 9.7% | – |
| 24 | GPT-5.6 Luna (xhigh) API access | OpenAI | 9.3% | – |
| 25 | Claude Opus 4.5 (high) API access | Anthropic | 7.1% | – |
| 26 | Inkling (xhigh) Open weights (unrestricted) | Thinking Machines | 4.4% | – |
