WeirdML (v2)
Science · 2025-06-27
Machine-learning engineering benchmark covering unusual or nonstandard ML tasks across a range of domains and implementation settings.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-6 Astra Pro | 93.6 |
| GPT-6 Astra | 93.3 |
| Claude Fable 5.1 | 92.9 |
| Claude Fable 5 | 91.9 |
| Claude Opus 5 | 91.8 |
| GPT-5.6 Sol Pro | 89.4 |
| GPT-5.6 Sol | 88.8 |
| GPT-5.5 | 84.9 |
| Gemini 3.8 Flash | 84.8 |
| Opus 4.8 | 82.9 |
| Kimi K3 | 82.6 |
| GPT-5.6 Terra | 78.3 |
| Opus 4.6 | 77.9 |
| GPT-5.3-Codex | 77.9 |
| GPT-5.4 | 77.7 |
| Opus 4.7 | 76.4 |
| GLM-5.3 | 75.4 |
| Qwen3.8 2.4T A95B | 75.2 |
| GPT-5.2 | 72.2 |
| Gemini 3.1 Pro Preview | 72.1 |
| GLM-5.2 | 70.1 |
| Gemini 3 Pro Preview | 69.9 |
| Sonnet 5 | 68.8 |
| Grok 4.6 | 67.3 |
| DeepSeek V4 Pro 0813 | 66.2 |
| Sonnet 4.6 | 66.1 |
| Opus 4.5 | 63.7 |
| DeepSeek-V4-Flash-0731 | 63.0 |
| Gemini 3.5 Flash | 62.6 |
| Gemini 3 Flash Preview | 61.6 |
| GPT-5.6 Luna | 60.9 |
| GPT-5.1 | 60.8 |
| GPT-5 | 60.7 |
| GPT-5 Pro | 60.4 |
| GPT-5.4 Mini | 60.3 |
| Muse Spark 1.2 | 60.3 |
| o3-pro | 58.2 |
| GLM-5.1 | 57.1 |
| Gemini 3.6 Flash | 56.1 |
| Kimi K2.6 | 55.9 |