Multi-turn task completion on IntentionGym
1.903Task Completion ScoreGRPO + Qwen2.5-8B
Evaluation Results
| Method | Links | |
|---|---|---|
| GRPO + Qwen2.5-8BCategory=RL (GRPO) Trained Models2026.04 | 1.903 | |
| GPT-4oCategory=Closed-Source LLM2026.04 | 1.898 | |
| Qwen2.5-32BCategory=Open-Source LLM2026.04 | 1.83 | |
| Qwen2.5-8B + PRIMECategory=PRIME Models2026.04 | 1.825 | |
| GRPO + Qwen2.5-4BCategory=RL (GRPO) Trained Models2026.04 | 1.808 | |
| Qwen2.5-14B + PRIMECategory=PRIME Models2026.04 | 1.782 | |
| Qwen2.5-4B + PRIMECategory=PRIME Models2026.04 | 1.772 | |
| Qwen2.5-4BCategory=Open-Source LLM2026.04 | 1.74 | |
| Qwen2.5-8BCategory=Open-Source LLM2026.04 | 1.725 | |
| Qwen2.5-14BCategory=Open-Source LLM2026.04 | 1.7 | |
| Gemini-1.5-FlashCategory=Closed-Source LLM2026.04 | 1.685 | |
| Gemini-1.5-ProCategory=Closed-Source LLM2026.04 | 1.59 | |
| GPT-4o-miniCategory=Closed-Source LLM2026.04 | 0.25 |