Why it matters
A mixture of Qwen 3 VL8B and Kimi K2.5 beat the state of the art on Video Web Arena, outperforming the leading GPT and Gemini models by 18 and 25 percent while costing 3.7 times less and running 3 times faster. The reason it worked is that visual web navigation decomposes into subtasks that do not all need a frontier m
My takeaway: A mixture of Qwen 3 VL8B and Kimi K2.5 beat the state of the art on Video Web Arena, outperforming the leading GPT and Gemini models by 18 and 25 percent while costing 3.7 times less and running 3 times faster. The reason it worked is that visual web navigation decomposes into subtasks that do not all need a frontier m