Qwen 3.7 Max

Qwen 3.7 Max is positioned by the Qwen team as an agent-first model rather than a single-turn answer engine, with a stated focus on staying coherent across extended workflows that involve planning, tool use, writing, and self-correction. Third-party reporting describes its design intent as covering agent-centric workloads such as coding and office productivity, where tasks are decomposed into multiple steps and refined iteratively. Rather than emphasizing raw prompt-and-response quality alone, the model is being showcased in contexts where an assistant must hold context together across many turns, decide when to act, and revise its own output as new information arrives.

In practical terms, this orientation makes the model a reasonable fit for building productivity agents and research assistants that need to gather context, draft artifacts, and validate results before returning a finished deliverable, rather than for short conversational snippets. Its emphasis on long-horizon execution aligns well with agent loops where the system interleaves reasoning, tool calls, and corrective passes over a shared working state. Developers choosing this model are effectively opting into a workflow-oriented design philosophy, trading single-shot responsiveness for sustained coherence across complex, tool-mediated tasks.

Vercel AI Gatewayalibaba/qwen3.7-maxqwen

Open provider page Provider docs

Quick Info

Provider: Vercel AI Gateway
Model key: alibaba/qwen3.7-max
Release date: May 21, 2026
Last updated: May 21, 2026
Input modalities
Output modalities
Capabilities

Cost

Input token cost: $2.50
Output token cost: $7.50

Limits

Output tokens: 64,000 tokens
Context window: 991,000 tokens

Transparent token rates

Compare Qwen 3.7 Max pricing

Rates are shown per one million tokens. Combined means one million input plus one million output tokens.

Browse this family

Latest news about Qwen 3.7 Max

Vercel AI Gateway

CoverageBenchmark

Qwen 3.7 Max was formally announced on May 20, 2026 at the Alibaba Cloud Summit in Hangzhou as Alibaba's proprietary, closed-weight flagship reasoning model, breaking from Alibaba's historical open-weight tradition. The release pairs a 1-million-token context window (up from 256K on Qwen 3.6 Max Preview) with default-o Pricing landed on OpenRouter at $2.50 per million input tokens and $7.50 per million output tokens, with a free preview available via chat.qwen.ai and lmarena.ai at launch; official Alibaba Cloud pricing was not public at the time of release. Alibaba claims a 35-hour continuous autonomous run executing more than 1,000

Vercel AI Gateway

Coverage

Qwen 3.7 Max-Preview launched May 19-20, 2026 at the Alibaba Cloud Summit with a native 1-million-token context window and 65,536-token maximum output, not a sliding-window approximation of long context. The model posts 90.4 on the MRCR-v2 128k retrieval benchmark, a score most competitors marketed at 1M context fail t API pricing is set at $2.50 per million input tokens, $7.50 per million output tokens, $0.50 per million cached reads, and $3.125 per million cache writes, though effective costs run 3–4× headline rates on long agent sessions because default extended thinking makes responses verbose. The model is pitched as a plug-repl

Vercel AI Gateway

CoverageBenchmark

Qwen 3.7 Max is Alibaba's closed-weight flagship reasoning model in the Qwen 3.7 family, announced May 20-21, 2026 at the Alibaba Cloud Summit in Hangzhou alongside the multimodal Qwen 3.7 Plus variant. It carries a 1-million-token context window and achieved an Intelligence Index v4.0 score of 56.6, ranking 5th overal An August 3, 2026 update on this page confirms that Qwen 3.8 Max has since launched as the successor with 2.4T total parameters, 95B active, Vision Arena 2, and a reported 16-day autonomous coding run, superseding the 3.7 Max tier. Qwen 3.7 Max remains text-only at the Max tier with vision handled by the separate Plus

Vercel AI Gateway

CoverageRelease Notes

Brief English reported on May 22, 2026 that Vercel officially expanded its AI Gateway model support by adding Qwen 3.7 Max to its unified API interface. This update allows developers using Next.js and SvelteKit to integrate the latest model while leveraging Vercel's centralized observability and security features. The integration includes specific metadata handling changes that may impact existing logging and monitoring workflows. Engineering teams are advised to update their Vercel AI SDK, configure AI Gateway provider settings, review security and rate limiting policies, audit application logging for PII leakage, and update en

Vercel AI Gateway

Official sourceRelease Notes

Vercel's first-party changelog announces that Qwen 3.7 Max from Alibaba is now available on Vercel AI Gateway (published May 21, 2026). The model is positioned as an agent foundation spanning coding, office workflow automation, and long-horizon autonomous execution, with stated improvements in frontend prototyping and The changelog frames the integration within AI Gateway's broader capabilities: a unified API for calling models, tracking usage and cost, and configuring retries, failover, and performance optimizations for higher-than-provider uptime. It also notes built-in custom reporting, observability, Bring Your Own Key support,