Benchmarks
331 sources · 1,258 models × 2,980 benchmarks · 155,643 scoresWorkflowArena's continuous Overall Score over the same 145 tasks: each task's verifier score from 0 to 100 credits partial progress (functional groups, required proofs, numerical targets, rubric items), is averaged within each of the 8 domains, and the domain means are averaged with equal weight. It runs well above the resolved rate, since most submissions earn partial credit while missing an acceptance criterion.
No signal or difficulty estimate. Too few models have been measured on this benchmark to estimate its difficulty.
| # | Model | Setting | Score | Source |
|---|---|---|---|---|
| 1 | Claude Opus 5.5 | high · Claude Code | 80.2 | |
| 2 | GPT-6 Astra | x-high · Codex | 77.7 | |
| 3 | GPT-6.1 Sol | x-high · Codex | 77.5 | |
| 4 | Claude Sonnet 5.5 | max · Claude Code | 73.1 | |
| 5 | Kimi K3.1 Pro | max · Claude Code | 62.7 | |
| 6 | GPT-6 Luna | max · Codex | 55.4 | |
| 7 | GLM-5.3 Flash | max · Claude Code | 39.0 |