Qwen 3.7 Max Launches on Vercel AI Gateway — Alibaba's 1M-Token Agent Model Sustained Autonomous Execution for 35 Hours

Vercel Blog·high signal

Alibaba announced Qwen3.7-Max at the 2026 Cloud Summit on May 20, and Vercel immediately added it to their AI Gateway. The proprietary model features a 1M-token context window and 65K max output, and in internal tests autonomously performed 1,000+ tool calls over 35 hours to optimize inference kernels, achieving a claimed 10x speedup. Priced at $2.50/$7.50 per 1M tokens on Model Studio, OpenRouter, and Together AI.