List your company

Web Navigation and Automation on WorkArena Held-out Tasks (test)

70Success RateClaude-3.5-Sonnet
1.56819.33437.154.866Jul 5, 2025

Evaluation results

MethodLinks
Claude-3.5-SonnetBackbone Model=Claude-3.5-Sonnet2025.0770
o1-MiniBackbone Model=o1-Mini2025.0768.6
Llama-3.1-405B-InstructBackbone Model=Llama-3.1-405B2025.0758.6
GPT-4oBackbone Model=GPT-4o2025.0755.7
Llama-3.3-70B Instruct (Teacher)Backbone Model=Llama-3.3-70B, Model Role=Teacher2025.0744
Llama-3.1-70B-InstructBackbone Model=Llama-3.1-70B2025.0732.9
Llama-3.1-8B SFT+RL (Ours)Backbone Model=Llama-3.1-8B, Training Protocol=SFT+RL2025.0728.8
GPT-4o-MiniBackbone Model=GPT-4o-Mini2025.0728.6
Qwen-2.5-72B Instruct (Teacher)Backbone Model=Qwen-2.5-72B, Model Role=Teacher2025.0727
Llama-3.1-8B SFT (Ours)Backbone Model=Llama-3.1-8B, Training Protocol=SFT2025.0726.4
Qwen-2.5-7B SFT+RL (Ours)Backbone Model=Qwen-2.5-7B, Training Protocol=SFT+RL2025.0725
Qwen-2.5-7B SFT (Ours)Backbone Model=Qwen-2.5-7B, Training Protocol=SFT2025.0721
Llama-3.1-8B RL (Ours)Backbone Model=Llama-3.1-8B, Training Protocol=RL2025.0711.5
Qwen-2.5-7B Instruct (Student)Backbone Model=Qwen-2.5-7B, Model Role=Student2025.0710
Qwen-2.5-7B RL (Ours)Backbone Model=Qwen-2.5-7B, Training Protocol=RL2025.077
Llama-3.1-8B Instruct (Student)Backbone Model=Llama-3.1-8B, Model Role=Student2025.074.2