Toolathlon — Same underlying benchmark as toolathlon but reports mean turn-count, a wholly different metric axis
| # | Model | Vendor | Best score | Runs | Last seen |
|---|---|---|---|---|---|
| 1 | Claude Sonnet 4.5 | Anthropic | 32 | 1 | 2026-06-30 |
| 2 | Claude Opus 4.7 | Anthropic | 25.9 | 1 | 2026-06-30 |
| 3 | Claude Opus 4.8 | Anthropic | 24.5 | 1 | 2026-06-30 |
| 4 | Claude Fable 5 | Anthropic | 19.8 | 1 | 2026-06-30 |
| 5 | Claude Mythos 5 | Anthropic | 17.6 | 1 | 2026-06-30 |
| 6 | Claude Opus 4.6 | Anthropic | 16.9 | 1 | 2026-06-30 |
| 7 | Claude Sonnet 4.6 | Anthropic | 16.5 | 1 | 2026-06-30 |
Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.