I'm finding some surprising performance on benchmarks from open weight models when compared to top models https://broskees.github.io/llm-compare/?m=claude-opus-

I'm finding some surprising performance on benchmarks from open weight models when compared to top models https://broskees.github.io/llm-compare/?m=claude-opus-4-6%2Cgpt-5.3-codex%2Ckimi-k2.5%2Cclaude-sonnet-4-6%2Cglm-5%2Cminimax-m2.5 https://broskees.github.io/llm-compare/?m=claude-opus-4-6%2C...