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

Safe Code Generation on Python Safe Code Generation Dataset

100Safe Code Rate

Claude Haiku 4.5

Updated 1mo ago

Evaluation Results

Method	Links
Claude Haiku 4.5 2026.06		100
Claude Sonnet 4.6 2026.06		98.59
GPT-5.4 2026.06		95.77
GPT-5.5 2026.06		94.96
Claude Haiku 4.5 2026.06		91.55
Claude Sonnet 4.6 2026.06		91.55
GPT-5.4 2026.06		83.8
GPT-5.4-mini 2026.06		81.69
DeepSeek V4 Pro 2026.06		80.28
GPT-5.5 2026.06		80.14
DeepSeek V4 Flash 2026.06		71.83
GPT-5.4-mini 2026.06		62.68
DeepSeek V4 Pro 2026.06		50
Claude Haiku 4.5 2026.06		34.51
GPT-5.4 2026.06		30.28
GPT-5.5 2026.06		26.62
DeepSeek V4 Flash 2026.06		25.35
Claude Sonnet 4.6 2026.06		11.97
DeepSeek V4 Pro 2026.06		7.75
GPT-5.4-mini 2026.06		7.04
DeepSeek V4 Flash 2026.06		5.63