TeamBench: leaderboard
Agent systems often decompose a task across multiple roles, but these roles are typically specified by prompts rather than enforced by access controls. Without enforcement, a team pass rate can mask whether agents actually coordinated or whether one role effectively did another role's work. We present TeamBench, a benchmark with 851 task templates and 931 seeded instances for evaluating agent coor
Metric: Solo (self-reported). Source: benchmarklist.com. Saturation forecast: Around September 2027. 13 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.7 | 35.6 | #49 |
| 2 | GPT-5.4 Mini | 33.3 | #212 |
| 3 | Gemini 3.1 Pro (Preview) | 27.8 | #58 |
| 4 | Gemma 4 31B | 27.8 | #200 |
| 5 | GPT-OSS-20B | 17.8 | #518 |
| 6 | Gemini 3 Flash | 13.3 | #99 |
| 7 | GPT-5.4 | 12.2 | #79 |
| 8 | Claude Haiku 4.5 | 12.2 | #280 |
| 9 | Claude Sonnet 4.6 | 7.8 | #90 |
| 10 | Gemini 3.1 Flash Lite | 5.6 | #191 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=teambench · How It Works · Data refreshed daily, snapshot 2026-10-20.