Video-Arena - Agentic Claude Code sc2: leaderboard
Video-Arena: 180 cases, 20 per game. Strict tuple F1 requires every non-temporal field to match and timestamps within 3 seconds; overall is the mean of nine game means. Paper v1 Appendix B.3 fixes high effort for OpenAI, Anthropic and Gemini; Kimi/MiniMax use unspecified provider defaults. Qwen thinking modes remain distinct. Full MP4, Claude Code harness, at most 200 turns and 120 minutes per case. Metric: Strict tuple F1 (%).
Metric: Strict tuple F1 (%). Source: videoarena-bench.github.io. Saturation forecast: Around 2029. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 5 (High) | 14.6 | #13 (Claude Opus 5) |
| 2 | Kimi K3 | 4.5 | #33 |
| 3 | MiniMax-M3 | 0.8 | #150 |
| 4 | Qwen 3.8 27B (Non-reasoning) | 0.5 | #112 (Qwen 3.8 27B) |
| 5 | Qwen 3.5 122B A10B | 0 | #179 |
No result here: #5 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=video-arena-agentic-claude-code-sc2 · How It Works · Data refreshed daily, snapshot 2026-10-26.