ClawArena-Team: leaderboard
Metric: Subagent-Management Score (%): task correctness multiplied by a least-privilege and modality-routing management factor in [0, 1]; 41 multi-turn multimodal scenarios with 258 evaluation rounds; a text-only main agent with partial workspace access creates, empowers and schedules a fixed locally served pool of LLM, VLM and omni subagents; execution-based checks, no LLM judge; higher is better. Source: arxiv.org. Saturation forecast: Around August 2027. 12 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Fable 5 | 60 |
| 2 | Gemini 3.5 Flash | 53.8 |
| 3 | GPT-5.5 | 51 |
| 4 | GLM-5.2 | 50.4 |
| 5 | GPT-5.4 | 49.7 |
| 6 | Gemini 3.1 Pro (Preview) | 49.6 |
| 7 | Kimi K2.6 | 49 |
| 8 | Claude Sonnet 4.6 | 48.5 |
| 9 | DeepSeek V4 Pro | 46.4 |
| 10 | Qwen 3.6 27B | 46.4 |
| 11 | Gemma 4 31B | 43.9 |
| 12 | GLM-4.7 Flash | 15.3 |
Interactive version: theaggregate.ai/benchmark?slug=clawarena-team · How It Works · Data refreshed daily, snapshot 2026-09-29.