Embodied Agent Instruction Following on ALFWorld (ALF) (test)
91Success RateEPPO
Evaluation Results
| Method | Links | |
|---|---|---|
| EPPOCategory=RL Training2026.07 | 91 | |
| AgentRLCategory=RL Training2026.07 | 89.2 | |
| AgentLM-70BCategory=Open LLMs, Parameters=70B2026.07 | 86 | |
| AgentLM-7BCategory=Open LLMs, Parameters=7B2026.07 | 84 | |
| BAPOCategory=RL Training2026.07 | 80.9 | |
| AgentLM-13BCategory=Open LLMs, Parameters=13B2026.07 | 76 | |
| Claude-Sonnet-4Category=API LLMs2026.07 | 73.6 | |
| Claude-Sonnet-4 ThinkingCategory=API LLMs, Variant=Thinking2026.07 | 69 | |
| GPT-5Category=API LLMs2026.07 | 65.4 | |
| DeepSeek-R1Category=Open LLMs2026.07 | 51.4 | |
| Hephaestus-8B-IFTCategory=Open LLMs, Parameters=8B, Variant=IFT2026.07 | 46 | |
| o4-miniCategory=API LLMs2026.07 | 32.6 | |
| Qwen2.5-32B-InstructCategory=Open LLMs, Parameters=32B, Variant=Instruct2026.07 | 32.1 | |
| DeepSeek-V3Category=Open LLMs2026.07 | 31.9 | |
| Hephaestus-8B-BaseCategory=Open LLMs, Parameters=8B, Variant=Base2026.07 | 30 | |
| o3-miniCategory=API LLMs2026.07 | 28.4 | |
| Qwen2.5-14B-InstructCategory=Open LLMs, Parameters=14B, Variant=Instruct2026.07 | 8.7 | |
| Qwen2.5-3B-InstructCategory=Open LLMs, Parameters=3B, Variant=Instruct2026.07 | 2.2 | |
| DCPOCategory=RL Training2026.07 | 0 |