Share your thoughts, 1 month free Claude Pro on usSee more

General AI Assistant Tasks on GAIA (avg@8)

88.5Avg@8 Score

MiroThinker-H1

Updated 1mo ago

Evaluation Results

Method	Links
MiroThinker-H1 2026.03		88.5
MiroThinker-1.7 2026.03		82.7
MiroThinker-v1.0-72B 2025.11		81.9
MiroThinker-1.7-mini 2026.03		80.3
OpenAI-GPT-5 2026.03		76.4
OpenAI-GPT-5-high 2025.11		76.4
Minimax-M2 2025.11		75.7
MiroThinker-v1.0-30B 2025.11		73.5
GLM-4.6 2025.11		71.9
Claude-4.5-Sonnet 2025.11		71.2
Tongyi-DeepResearch-30B 2026.03		70.9
Tongyi-DeepResearch-30B 2025.11		70.9
Claude-4-Sonnet 2025.11		68.3
OpenAI DeepResearch 2025.11		67.4
MiroThinker-v1.0-8B 2025.11		66.4
SFR-DeepResearch-20B 2025.11		66
DeepSeek-V3.2 2025.11		63.5
DeepSeek-V3.1 2025.11		63.1
Kimi-K2-0905 2025.11		60.2
DeepMiner-32B-RL 2025.11		58.7
AFM-32B-RL 2025.11		55.3
WebExplorer-8B-RL 2025.11		50