List your company

Multi-turn task completion on TurtleGym

29.2Task Completion ScoreGPT-4o
7.67213.26118.8524.439Apr 8, 2026

Evaluation results

MethodLinks
GPT-4oCategory=Closed-Source LLM2026.0429.2
Gemini-1.5-ProCategory=Closed-Source LLM2026.0427.4
Gemini-1.5-FlashCategory=Closed-Source LLM2026.0419.6
GRPO + Qwen2.5-8BCategory=RL (GRPO) Trained Models2026.0419.2
GRPO + Qwen2.5-4BCategory=RL (GRPO) Trained Models2026.0418.4
Qwen2.5-14B + PRIMECategory=PRIME Models2026.0416.8
Qwen2.5-32BCategory=Open-Source LLM2026.0415.1
Qwen2.5-14BCategory=Open-Source LLM2026.0414.2
Qwen2.5-8B + PRIMECategory=PRIME Models2026.0414.2
Qwen2.5-4B + PRIMECategory=PRIME Models2026.0413.2
Qwen2.5-8BCategory=Open-Source LLM2026.049.8
GPT-4o-miniCategory=Closed-Source LLM2026.049.1
Qwen2.5-4BCategory=Open-Source LLM2026.048.5