Regular display scope. Rankings are grouped by metric and evaluation protocol. Recorded tool access, budgets, and frameworks are shown separately. Scores remain available for reference when conditions are unspecified.
win rate
Tool access not recorded / Output limit 32768 tokens / Framework not recorded · Thinking MoE · max_output 32768
Arena-Hard V2 · Qwen3-VL技术报告v2;Table 6; page 20 footnote says win rates evaluated by GPT-4.1; Thinking MoE sampling settings · win rate · Tool access not recorded / Output limit 32768 tokens / Framework not recorded · Thinking MoE · max_output 32768