LLM-LIBERO - LIBERO-10: leaderboard
LLM-LIBERO: 100 episodes per configuration, 10 long-horizon robot manipulation tasks; visual observations, proprioception and structured pose/gripper tool calls, no task-specific fine-tuning. Six frontier models at Low, Medium and High requested reasoning, 1,800 episodes total. Timeout/failure retained in denominator; cost and call counts omitted. Success Rate (%).
Metric: Success Rate (%). Source: arxiv.org. Saturation forecast: Around December 2027. 18 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-6 Astra (High) | 44 | #6 (GPT-6 Astra) |
| 2 | GPT-6 Astra (Medium) | 38 | #6 (GPT-6 Astra) |
| 3 | GPT-6.1 Sol (High) | 33 | #7 (GPT-6.1 Sol) |
| 4 | Claude Opus 5.5 (High) | 33 | #4 (Claude Opus 5.5) |
| 5 | GPT-6.1 Sol (Medium) | 30 | #7 (GPT-6.1 Sol) |
| 6 | GPT-6 Astra (Low) | 28 | #6 (GPT-6 Astra) |
| 7 | GPT-6.1 Sol (Low) | 24 | #7 (GPT-6.1 Sol) |
| 8 | Claude Opus 5.5 (Medium) | 18 | #4 (Claude Opus 5.5) |
| 9 | Claude Opus 5 (Medium) | 12 | #12 (Claude Opus 5) |
| 10 | Claude Opus 5 (Low) | 10 | #12 (Claude Opus 5) |
| 11 | Claude Opus 5.5 (Low) | 9 | #4 (Claude Opus 5.5) |
| 12 | Claude Opus 5 (High) | 6 | #12 (Claude Opus 5) |
| 13 | GPT-6 Luna (Medium) | 0 | #60 (GPT-6 Luna) |
| 14 | GPT-6 Luna (High) | 0 | #60 (GPT-6 Luna) |
| 15 | GPT-6 Luna (Low) | 0 | #60 (GPT-6 Luna) |
No result here: #8 Gemini 4 Argon, #9 Claude Fable 5.1, #10 Muse Wonder Full.
Interactive version: theaggregate.ai/benchmark?slug=llm-libero-libero-10 · How It Works · Data refreshed daily, snapshot 2026-10-23.