EditionUpdated daily
The reference ranking
of artificial intelligence.
Results from the toughest benchmarks, combined into a single score. Every number is sourced.
Overall ranking
ERAZ points out of 1000. Click a model to see its detailed results.
| # | Model | ERAZ points | Reas. | Math | Code | Agents | #1 | Cov. |
|---|
"—" = result not published or not verified by an independent source. The ranking can also be to your priorities.
Model pages
- 1GPT-6 Astra
- 2Claude Opus 5.5
- 3Claude Opus 5
- 4Claude Fable 5.1
- 5GPT-6.1 Sol
- 6Gemini 4 Argon
- 7Claude Sonnet 5.5
- 8GPT-5.6 Sol
- 9Muse Wonder Full
- 10GPT-5.6 Terra
- 11Muse Wonder
- 12Gemini 3.8 Flash
- 13Kimi K3
- 14Step 5 Preview
- 15GLM-5.3
- 16MiniMax M3.1 Flash
- 17DeepSeek V4.1 Flash
- 18Qwen3.8-Flash-Next
- 19MiMo-V2.6-Pro
- 20Grok 4.7
- 21Qwen3.8 Max
- 22GLM-5.3-Flash
- 23Claude Haiku 5.5
- 24Ling 3.1 Flash
- 25Qwen3.8 2.4T A95B
- 26Mistral Large 4