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

Scientific Reasoning on SuperGPQA (test)

77.7Pass@1

GPT-5.4-xhigh

Updated 12d ago

Evaluation Results

Method	Links
GPT-5.4-xhigh 2026.05		77.7
DeepSeek-V3.2 2026.05		75
MiroThinker-v1.5 2026.05		73.7
GPT-5.4-xhigh 2026.05		73.5
SR2AM-v1.0-30B 2026.05		73
Kimi-K2.5 2026.05		72.5
DeepSeek-V3.2 2026.05		71.8
GLM-4.6 2026.05		67
Qwen3-235B 2026.05		66.3
ASearcher-Web-QWQ-v2 2026.05		65
Qwen3-30B-A3B 2026.05		63.1
Qwen3-30B-A3B 2026.05		62.4
WebExplorer-8B 2026.05		60.3
GPT-OSS-120B-high 2026.05		58.6
K2-Think-V2-high 2026.05		58
SR2AM-v0.1-8B 2026.05		58
A2FM 2026.05		55.3
Tongyi-DeepResearch 2026.05		54.9
SimpleTIR-32B 2026.05		50.5
WebSailor-32B 2026.05		47.3
SimpleTIR-7B 2026.05		41.7
ASearcher-Web-7B 2026.05		35.7
AFM-Web-7B-RL 2026.05		32.4
WebSailor-7B 2026.05		28
AFM-Code-7B-RL 2026.05		23.6