Qwen 3.5 Max Preview Breaks Into Arena Expert Top 10

Mar 20, 2026 · Updated Apr 25, 2026

Qwen 3.5 Max Preview, Alibaba's latest large language model, has debuted on the Arena.ai leaderboard with a score of 1498 on Arena Expert prompts — landing at #10 overall. The model ranks #3 in Math on Expert prompts and #5 in Math on Text Arena, with additional top-20 placements across Writing, Literature & Language, Life Sciences, and Social Science. The result carries a Preliminary flag, indicating rankings based on early vote counts.

Arena.ai runs independent human preference evaluations where real users compare model outputs blind. Landing in the Expert top 10 — which filters for harder, domain-specific prompts — places it among established frontier models on what has become a widely referenced benchmark for model quality.

Filter the Arena leaderboard by Math or Expert Prompts to compare qwen3.5-max-preview against the models you're evaluating.

View the full update on arena.ai

Arena.ai

@arenaMar 19

Qwen 3.5 Max Preview has landed in top 10 for Arena Expert and top 15 for Text Arena. It shows particular strength in Math. Highlights: - #3 Math - #10 Expert - #15 Text Arena - Top 20 for Writing, Literature & Language, Life, Physical, & Social Science, Entertainment, Sports, https://t.co/VxvbSkNRqC

View on X