ResearchBenchmarksWeb Navigation on Multimodal-Mind2WebFollow48.49Success RateCUA37.965240.697643.4346.1624Mar 23, 2026Evaluation ResultsMethodMethodLinksSuccess RateCUAOrganization=OpenAIOrganization=OpenAI2026.0348.49ReCAP-32BBackbone=Qwen3-VL-Thin...Backbone=Qwen3-VL-Thinking-32B, Status=Finetuned2026.0344.73Qwen-32BBackbone=Qwen3-VL-Thin...Backbone=Qwen3-VL-Thinking-32B2026.0342.91Qwen-8BBackbone=Qwen3-VL-Thin...Backbone=Qwen3-VL-Thinking-8B2026.0342.58ReCAP-8BBackbone=Qwen3-VL-Thin...Backbone=Qwen3-VL-Thinking-8B, Status=Finetuned2026.0338.37