I transposed the table so that it's readable on mobile devices.
I should have mentioned that the Qwen 3.5 benchmarks were from the Qwen3.5-122B-A10B model card (which includes GPT-5-mini and GPT-OSS-120B); apologies for not including the smaller Qwen 3.5 models.