CLEM TextMapWorld SpecificRoom — leaderboard

Metric: Game Clemscore (%). Source: huggingface.co. 31 models tracked.

Top models

#ModelScore
1GPT-5.2100
2Llama 3.1 70B Instruct100
3Claude Sonnet 4.5100
4Kimi K2 (Thinking)100
5GLM-4.6100
6GPT-5.2 (Medium)100
7Gemini 3 Flash100
8GPT-5.2 (High)100
9GLM-5 (Non-reasoning)100
10Claude Sonnet 4.5 (High)100
11Qwen 2.5 72B Instruct96.67
12DeepSeek V3.296.67
13Mistral Large 396.67
14GLM-4.796.67
15Qwen 3 Max96.67

Interactive version: theaggregate.ai/benchmark?slug=clem-textmapworld-specificroom · How the rankings work · Data refreshed daily, snapshot 2026-07-22.