InteractiveBench Planning (task pass)
Agents · 2026-10-25
InteractiveBench's Planning category: the 20 Planning tasks, in which the assistant plans and re-plans work under time and resource limits (for example cooking safe dishes for guests with allergies with one oven and pan in 30 minutes). Same simulator, Gemini 3.1 Pro Human agent and end-state check as the full benchmark; the score is the percentage of these 20 tasks in which every required end state holds.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Fable 5.1 | 25.0 |
| Claude Opus 5 | 25.0 |
| GPT-5.6 Sol | 25.0 |
| Muse Spark 1.1 | 25.0 |
| Opus 4.8 | 20.0 |
| GLM-5.3 Flash | 20.0 |
| GPT-5.5 | 15.0 |
| Kimi K3 | 15.0 |
| Gemini 3.1 Pro Preview | 10.0 |
| GPT-5.6 Luna | 5.0 |