TheoPhysEval benchmark
GPT-6-Astra solved 49 out of 50 problems on the TheoPhysEval benchmark Anthropic Opus 5.5 solved 92% of the problems on the TheoPhysEval benchmark Google Gemini 3.8 Flash solved 68% of the problems on the TheoPhysEval benchmark DeepSeek-V4.1-Flash solved 54% of the problems on the TheoPhysEval benchmark GLM 5.3 solved 42% of the problems on the TheoPhysEval benchmark The TheoPhysEval benchmark is available on the Hugging Face dataset· Graphiti 生成的实体摘要
关系变化
按生效时间- EvalEval CoalitionSOLVEDTheoPhysEval benchmark
Anthropic Opus 5.5 solved 92% of the problems on the TheoPhysEval benchmark
- GLM-5.3SOLVEDTheoPhysEval benchmark
GLM 5.3 solved 42% of the problems on the TheoPhysEval benchmark
- SWE-bench VerifiedSOLVEDTheoPhysEval benchmark
Google Gemini 3.8 Flash solved 68% of the problems on the TheoPhysEval benchmark
- GPT-6 AstraSOLVEDTheoPhysEval benchmark
GPT-6-Astra solved 49 out of 50 problems on the TheoPhysEval benchmark
- CV3 EvalSOLVEDTheoPhysEval benchmark
DeepSeek-V4.1-Flash solved 54% of the problems on the TheoPhysEval benchmark
- TheoPhysEval benchmarkAVAILABLE ONIndirect Prompt Injection (IPI) benchmark
The TheoPhysEval benchmark is available on the Hugging Face dataset
关系是 Graphiti 从报道中提取的事实陈述，不是原文引述；失效时间表示后来的报道显示这条关系已不成立。