SMH-Bench (Direct Reasoning) - Automation Scheduling: leaderboard

Metric: Success rate (%) on Automated Task Scheduling (TC4, 150 time- or state-triggered rules, trigger and action both verified); Direct Reasoning: the full serialized home state (rooms, devices, current values, service schemas) and the request in one prompt, one structured JSON answer replayed in the HomeEnv simulator; final home state checked against the goal with untouched devices preserved; clarification and query answers judged by GPT-5; higher is better. Source: arxiv.org. Saturation forecast: Around December 2026. 13 models tracked.

Top models

#ModelScore
1Gemini 3.1 Pro (Preview)70.7
2Claude Sonnet 4.668.7
3GPT-5.465.3
4Claude Haiku 4.564
5Qwen 3.5 397B A17B60
6GLM-552
7DeepSeek V3.2 (Thinking)52
8Qwen 3.5 Plus50
9GPT-5.4 Mini45.3
10MiniMax-M2.738.7
11Qwen 3.5 9B25.3
12Qwen 3.5 4B24

Interactive version: theaggregate.ai/benchmark?slug=smh-bench-direct-reasoning-automation-scheduling · How It Works · Data refreshed daily, snapshot 2026-09-29.