How well the model recovers after a failed terminal command instead of thrashing. Ranked by Arena agent_bash_recovery_steps, priced against live API rates. 30 models have a published score on this board.
Capability published 2026-10-08 · pricing retrieved 2026-10-25
anthropic/claude-sonnet-5.5anthropic/claude-sonnet-5.5anthropic/claude-sonnet-5.5anthropic/claude-sonnet-5.5Rows marked frontier are Pareto-optimal — nothing we track is both stronger and cheaper. "Strength" rescales the board so 100% = parity with the leader.
| # | Model | Provider | Score | Strength | $/1M | Context | Tools |
|---|---|---|---|---|---|---|---|
| 1 | Claude Sonnet 5.5anthropic/claude-sonnet-5.5 | Anthropic | 0.1504 | 100% | $4.00 | 1M | tools |
| 2 | Claude Opus 5anthropic/claude-opus-5 | Anthropic | 0.1357 | 90% | $10.00 | 1M | tools |
| 3 | Claude Opus 5.5anthropic/claude-opus-5.5 | Anthropic | 0.1355 | 90% | $8.00 | 1M | tools |
| 4 | Claude Fable 5.1anthropic/claude-fable-5.1 | Anthropic | 0.1160 | 77% | $20.00 | 1M | tools |
| 5 | Claude Fable 5anthropic/claude-fable-5 | Anthropic | 0.0960 | 64% | $20.00 | 1M | tools |
| 6 | DeepSeek V4.1 Flash 1017deepseek/deepseek-v4.1-flash-1017 | DeepSeek | 0.0786 | 52% | $0.26 | 1.048576M | tools |
| 7 | DeepSeek V4.1 Flashdeepseek/deepseek-v4.1-flash | DeepSeek | 0.0786 | 52% | $0.27 | 1.048576M | tools |
| 8 | Claude Sonnet 5anthropic/claude-sonnet-5 | Anthropic | 0.0738 | 49% | $4.00 | 1M | tools |
| 9 | Claude Opus 4.8anthropic/claude-opus-4.8 | Anthropic | 0.0722 | 48% | $10.00 | 1M | tools |
| 10 | Grok 4.7x-ai/grok-4.7 | xAI | 0.0720 | 48% | $3.00 | 500K | tools |
| 11 | GPT-6 Solopenai/gpt-6-sol | OpenAI | 0.0703 | 47% | $4.00 | 1.05M | tools |
| 12 | Grok 4.6x-ai/grok-4.6 | xAI | 0.0568 | 38% | $3.00 | 500K | tools |
| 13 | GPT-6 Astraopenai/gpt-6-astra | OpenAI | 0.0553 | 37% | $20.00 | 1.05M | tools |
| 14 | GPT-6 Lunaopenai/gpt-6-luna | OpenAI | 0.0521 | 35% | $0.20 | 1.05M | tools |
| 15 | Grok 4.5x-ai/grok-4.5 | xAI | 0.0514 | 34% | $3.00 | 500K | tools |
| 16 | GPT-6.1 Solopenai/gpt-6.1-sol | OpenAI | 0.0453 | 30% | $4.00 | 1.05M | tools |
| 17 | DeepSeek V4 Pro 0813deepseek/deepseek-v4-pro-0813 | DeepSeek | 0.0440 | 29% | $0.99 | 1.048576M | tools |
| 18 | DeepSeek V4 Pro 0423deepseek/deepseek-v4-pro | DeepSeek | 0.0440 | 29% | $0.26 | 1.048576M | tools |
| 19 | Kimi K3moonshotai/kimi-k3 | Moonshot AI | 0.0438 | 29% | $2.88 | 1.048576M | tools |
| 20 | Qwen3.8 Max (0902)qwen/qwen3.8-max-0902 | Qwen (Alibaba) | 0.0280 | 19% | $3.00 | 1M | tools |
| 21 | GLM 5.2z-ai/glm-5.2 | Z.ai (Zhipu) | 0.0067 | 4% | $1.77 | 1.048576M | tools |
| 22 | MiniMax M3minimax/minimax-m3 | MiniMax | 0.0058 | 4% | $0.53 | 1.048576M | tools |
| 23 | GLM 5.3z-ai/glm-5.3 | Z.ai (Zhipu) | -0.0005 | 0% | $0.41 | 1.048576M | tools |
| 24 | Gemini 3.8 Flashgoogle/gemini-3.8-flash | -0.0069 | -5% | $1.66 | 1.048576M | tools | |
| 25 | Gemini 3.7 Flashgoogle/gemini-3.7-flash | -0.0544 | -36% | $1.50 | 1.048576M | tools |
Capability: Arena leaderboard dataset (CC-BY-4.0), Arena agent_bash_recovery_steps, published 2026-10-08. Pricing: OpenRouter, retrieved 2026-10-25. (input x 3 + output x 1) / 4.
curl -s https://agentleaderboards.com/api/v1/route/agent-recovery.json | jq .picksNo key, open CORS, refreshed daily. Set your own budget and constraints →
Writing & editing code · Autonomous agent work · Tool-call reliability · Following operator instructions · Hard reasoning · Mathematics · Precise instruction following · Long multi-turn chat · Long inputs · Creative writing · General assistant · Software & IT services · Legal & government · Medicine & healthcare · Science · Business & finance · Writing & language