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

Scientific Reasoning on HLE (test)

49Pass@1

GPT-5.4-xhigh

Updated 12d ago

Evaluation Results

Method	Links
GPT-5.4-xhigh 2026.05		49
GPT-5.4-xhigh 2026.05		40.8
DeepSeek-V3.2 2026.05		39.6
Kimi-K2.5 2026.05		34.4
Tongyi-DeepResearch 2026.05		31.6
SR2AM-v1.0-30B 2026.05		30.4
MiroThinker-v1.5 2026.05		30
DeepSeek-V3.2 2026.05		24
GPT-OSS-120B-high 2026.05		20.4
GLM-4.6 2026.05		19
ASearcher-Web-QWQ-v2 2026.05		16.4
WebExplorer-8B 2026.05		14.8
SR2AM-v0.1-8B 2026.05		14.2
WebSailor-32B 2026.05		12.2
Qwen3-235B 2026.05		12
K2-Think-V2-high 2026.05		11.6
A2FM 2026.05		10.3
Qwen3-30B-A3B 2026.05		10.2
Qwen3-30B-A3B 2026.05		9.8
WebSailor-7B 2026.05		9.4
AFM-Web-7B-RL 2026.05		7.8
ASearcher-Web-7B 2026.05		6.8
SimpleTIR-7B 2026.05		4.6
SimpleTIR-32B 2026.05		3.6
AFM-Code-7B-RL 2026.05		2.8