Alibaba unveils Qwen 3.7 Max — Intelligence Index 56.6, 97.1 HMMT 2026 February, closed weights on flagship for the first time
Alibaba launched Qwen 3.7 Max at the Alibaba Cloud Summit on May 20 with an Intelligence Index of 56.6 and 97.1 on HMMT 2026 February. The release is structurally significant: it's the first time Alibaba has kept the flagship Qwen variant closed-weight after years of shipping open-weight flagships. The shift mirrors the closed-lab strategic playbook and reframes Qwen's open-weight reputation.
The Intelligence Index score of 56.6 puts Qwen 3.7 Max within striking distance of the April 2026 frontier ceiling (GPT-5.5 xhigh at 60.24). The 97.1 HMMT score is the more interesting capability signal — it's a math benchmark that frontier closed models have only recently cleared. Qwen 3.7 Max landing in that performance neighborhood while keeping the weights closed represents Alibaba accepting the closed-weight strategic trade-off they previously rejected.
The market-positioning consequence is sharper. Qwen 3.6 Max-Preview (released April 20) holds #1 on six coding and agent benchmarks as an open-weight release. Qwen 3.7 Max ships as closed-weight. The two-track strategy — open-weight at one capability tier, closed-weight at the highest — directly mirrors the OpenAI / Anthropic / Google split between API-only flagships and lower-tier offerings. The Chinese open-source-leader brand is being repositioned in real time.
WhatLLM — New AI Models May 2026: The Frontier Took a Breath, Architecture Took the Stage → · LLM-Stats — AI Updates Today May 2026 Latest AI Model Releases → · FelloAI — Best AI Models in May 2026: ChatGPT, Claude, Gemini & Grok →