mcp Tasks

5 models evaluated

#	Model	Provider	Type	Score
1	Qwen3.5 397B	Alibaba	Open weights	74.2%
2	Qwen3.6 Plus	Alibaba	Proprietary	74.1%
3	Claude Opus 4.5	Anthropic	Proprietary	71.8%
4	GLM-5	Z.AI · Open weight	60.8%
5	Kimi K2.5	Moonshot AI · Open weight	59.1%