Agent Arena - Steerability: leaderboard
Agent Arena submetric measuring how well models adapt to user steering during tool-use sessions.
Metric: Steerability (%). Source: arena.ai. Saturation forecast: Around January 2028. 51 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-6 Sol (Max) | 11.92 | #14 (GPT-6 Sol) |
| 2 | Gemini 4 Argon (High) | 11.64 | #8 (Gemini 4 Argon) |
| 3 | Claude Fable 5 (High) | 9.49 | #16 (Claude Fable 5) |
| 4 | GPT-6 Astra (Max) | 8.85 | #6 (GPT-6 Astra) |
| 5 | Claude Fable 5.1 (Max) | 8.84 | #9 (Claude Fable 5.1) |
| 6 | GPT-6.1 Sol (Max) | 8.52 | #7 (GPT-6.1 Sol) |
| 7 | Claude Opus 4.8 (High) | 8.41 | #37 (Claude Opus 4.8) |
| 8 | Claude Opus 5.5 (Max) | 8.2 | #3 (Claude Opus 5.5) |
| 9 | GPT-5.4 (High) | 6.5 | #79 (GPT-5.4) |
| 10 | GPT-5.6 Sol (xHigh) | 6.46 | #19 (GPT-5.6 Sol) |
| 11 | Claude Sonnet 5.5 (Max) | 5.1 | #11 (Claude Sonnet 5.5) |
| 12 | Muse Spark 1.3 (Max) | 4.82 | #24 (Muse Spark 1.3) |
| 13 | Claude Sonnet 5 (High) | 4.57 | #57 (Claude Sonnet 5) |
| 14 | Claude Opus 5 (Max) | 4.22 | #12 (Claude Opus 5) |
| 15 | GPT-5.5 (xHigh) | 4.08 | #31 (GPT-5.5) |
No result here: #10 Muse Wonder Full.
Interactive version: theaggregate.ai/benchmark?slug=agent-arena-steerability · How It Works · Data refreshed daily, snapshot 2026-10-20.