WildAWM OpenCode - OSWorld-2: leaderboard
WildAWM (2026): common OpenCode v1.17.7 agent harness, high reasoning effort, eight public web, claw and computer-use evaluations. Fine-tuned WildAWM-4B and WildAWM-35B-A3B remain distinct from Qwen baselines; cited OpenForgeRL and CUA-Gym rows use different harnesses and are not pooled into this protocol. Mean Score (%); higher is better.
Metric: Mean Score (%). Source: arxiv.org. Saturation forecast: Around 2029. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-6 Astra (High) | 60.77 | #6 (GPT-6 Astra) |
| 2 | Claude Opus 5 (High) | 52.79 | #12 (Claude Opus 5) |
| 3 | Claude Opus 4.8 (High) | 48.32 | #37 (Claude Opus 4.8) |
| 4 | GPT-5.5 (High) | 47.12 | #30 (GPT-5.5) |
| 5 | GPT-5.6 Sol (High) | 46.89 | #18 (GPT-5.6 Sol) |
| 6 | GLM-5.2 (High) | 45 | #73 (GLM-5.2) |
No result here: #4 Claude Opus 5.5, #7 Gemini 4 Argon, #8 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=wildawm-opencode-osworld-2 · How It Works · Data refreshed daily, snapshot 2026-10-21.