Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Web Navigation Agentic Reasoning on BrowseComp complete (test)
Loading...
88.2
Avg Success Rate@3
MiroThinker-H1
41.608
53.704
65.8
77.896
Mar 16, 2026
Avg Success Rate@3
Updated 1mo ago
Evaluation Results
Method
Method
Links
Avg Success Rate@3
MiroThinker-H1
Agent Style=ReAct, Max...
2026.03
88.2
Gemini-3.1-Pro
Agent Style=ReAct, Max...
2026.03
85.9
Claude-4.6-Opus
Agent Style=ReAct, Max...
2026.03
84
OpenAI-GPT-5.4
Agent Style=ReAct, Max...
2026.03
82.7
Qwen3.5-397B
Agent Style=ReAct, Max...
2026.03
78.6
Kimi-K2.5
Agent Style=ReAct, Max...
2026.03
78.4
Seed-2.0-Pro
Agent Style=ReAct, Max...
2026.03
77.3
Minimax-M2.5
Agent Style=ReAct, Max...
2026.03
76.3
GLM-5.0
Agent Style=ReAct, Max...
2026.03
75.9
MiroThinker-1.7
Agent Style=ReAct, Max...
2026.03
74
MiroThinker-1.7-mini
Agent Style=ReAct, Max...
2026.03
67.9
Claude-4.5-Opus
Agent Style=ReAct, Max...
2026.03
67.8
DeepSeek-V3.2
Agent Style=ReAct, Max...
2026.03
67.6
Gemini-3.0-Pro
Agent Style=ReAct, Max...
2026.03
59.2
OpenAI-GPT-5
Agent Style=ReAct, Max...
2026.03
54.9
Tongyi-DeepResearch-30B
Agent Style=ReAct, Max...
2026.03
43.4
Feedback
Search any
task
Search any
task