Google Carbon Report - Agent Collaboration Eval 2.0 - Active Collaboration: leaderboard
Google DeepMind Carbon safety report, Figure 5.8. Ambiguous/high-risk SWE and non-coding tasks: correctly defer to user; keep distinct from autonomous execution on unambiguous tasks. Task count not stated. Metric: Pass@1 (%). Higher is better. Source: https://storage.googleapis.com/deepmind-media/gemini/gemini_4_carbon_safety_report.pdf#page=70
Metric: Pass@1 (%). Source: storage.googleapis.com. Saturation forecast: Around February 2027. 3 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 4 Carbon | 80 | #2 |
| 2 | Gemini 4 Argon | 70 | #11 |
| 3 | Gemini 3.8 Flash | 34 | #36 |
No result here: #1 Claude Fable 5.2, #5 Claude Opus 5.5, #6 GPT-6 Astra.
Interactive version: theaggregate.ai/benchmark?slug=google-carbon-report-agent-collaboration-eval-2-0-active-collaboration · How It Works · Data refreshed daily, snapshot 2026-11-12.