Claude Fable 5.264.9%
Anthropic internal simulationeffort: highest available · Internal simulation run; highest available reasoning effort.
Anthropic ↗Best LLMs for simulated robotics tasks
RoboDojo SimTests controlling a simulated robot from visual information, helping you assess progress toward automating physical tasks. These are internal simulation results, not evidence of reliable operation in a client's workplace.
Updated: Data as of:
| Model / creator | Version | Reported score |
|---|---|---|
 Claude Fable 5.2Anthropic · Proprietary | Anthropic internal simulation | 64.9% |
| Claude Opus 5.5Anthropic · Proprietary | Anthropic internal simulation | 39.4% |
 GPT-6 AstraOpenAI · Proprietary | Anthropic internal simulation | 29.8% |
| Claude Fable 5.1Anthropic · Proprietary | Anthropic internal simulation | 27.6% |
View all 4 published reports
Claude Opus 5.539.4%
Anthropic internal simulationeffort: highest available · November 9 launch-table comparison; highest available effort unless explicitly noted. Preserve the benchmark track and model configuration; competitor figures are as reported by Anthropic.
Anthropic ↗GPT-6 Astra29.8%
Anthropic internal simulationeffort: highest available · November 9 launch-table comparison; highest available effort unless explicitly noted. Preserve the benchmark track and model configuration; competitor figures are as reported by Anthropic.
Anthropic ↗Claude Fable 5.127.6%
Anthropic internal simulationeffort: highest available · November 9 launch-table comparison; highest available effort unless explicitly noted. Preserve the benchmark track and model configuration; competitor figures are as reported by Anthropic.
Anthropic ↗