live Code Bench Pro

6 models evaluated

#	Model	Provider	Type	Score
1	GPT-5.4	OpenAI	Closed	87.5%
2	Gemini 3.1 Pro	Google	Closed	82.9%
3	Muse Spark	Meta	Closed	80.0%
4	Grok 4.20	xAI	Closed	74.2%
5	Claude Opus 4.6	Anthropic	Closed	70.7%
6	MiniCPM5-1B	OpenBMB	Open	22.7%