AI capability frontier
An interactive atlas of LLM benchmarks: model scores, leaderboards, latent capability estimates, and dataset coverage across frontier AI models.
Top models by capability
| Model | Capability |
|---|---|
| Claude Opus 5.5 | 168.3 |
| GPT-6 Astra | 167.9 |
| Gemini 4 Argon | 167.3 |
| Claude Sonnet 5.5 | 167.3 |
| GPT-6.1 Sol | 166.9 |
| Claude Fable 5.1 | 166.7 |
| Kimi K3.1 Pro | 165.9 |
| Muse Wonder Full | 165.5 |
| Claude Mythos Preview | 165.2 |
| Claude Opus 5 | 164.9 |
| Claude Fable 5 | 164.7 |
| GPT-5.6 Sol Pro | 164.4 |
| Muse Wonder | 164.4 |
| GPT-6 Sol | 164.3 |
| Gemini 3 Deep Think | 164.0 |
| GPT-5.5 Pro | 163.9 |
| GPT-5.6 Sol | 163.4 |
| GPT-5.6 Terra Pro | 162.6 |
| DeepSeek V4.1 Flash 1017 | 162.3 |
| Fugu Ultra v1.0 | 162.2 |