STT-Arena - Hard: leaderboard
Hard tasks, long-horizon cascading constraints that need global replanning, of STT-Arena (2026), a tool-use benchmark in executable stateful environments whose conditions change in time and space during the task, with a simulated user. Pass@1 in percent (every rule-based check must hold in the final state), mean of three runs; higher is better.
Metric: Pass@1 (%) on the 59 hard tasks (long-horizon cascading constraints requiring global replanning); a task passes when every rule-based check function holds in the final state; the model acts in an executable stateful environment whose state changes under scheduled spatio-temporal triggers, with a passive Qwen3.5-397B user simulator and at most 50 turns; mean of three runs at temperature 0.7; higher is better. Source: arxiv.org. Saturation forecast: Around October 2028. 21 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.4 | 24.29 | #135 |
| 2 | Claude Opus 4.6 | 23.73 | #113 |
| 3 | Qwen 3.6 Plus | 23.16 | #238 |
| 4 | Gemini 3.1 Pro (Preview) | 21.47 | #99 |
| 5 | Claude Sonnet 4.6 | 19.77 | #173 |
| 6 | DeepSeek V3.2 | 15.82 | #419 |
| 7 | DeepSeek V4 Pro | 15.25 | #196 |
| 8 | GLM-5.1 | 15.25 | #198 |
| 9 | Kimi K2.5 | 14.12 | #266 |
| 10 | STT-Agent-4B (SFT+RL) | 14.12 | #316 |
| 11 | MiniMax-M2.7 | 13.56 | #539 |
| 12 | STT-Agent-4B (SFT) | 13.56 | #418 |
| 13 | Gemini 2.5 Flash | 12.43 | #463 |
| 14 | MiniMax-M2.5 | 12.43 | #600 |
| 15 | Qwen 3.5 9B | 11.3 | #828 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=stt-arena-hard · How It Works · Data refreshed daily, snapshot 2026-11-21.