Qwen3.7-max is the text-only reasoning flagship of Alibaba's Qwen 3.7 generation, announced at the Alibaba Cloud Summit on May 20, 2026. It is a proprietary, closed-weight model — there is no open-source qwen3.7-max release — currently available as a preview through the qwen3.7-max API. Vision and image input live in its sibling model, Qwen3.7-Plus; Max takes text in and returns text out. On GPTProto, you reach the qwen3.7-max API on the same balance and key as every other model on the platform.
| Spec | qwen3.7-max |
|---|---|
| Provider | Alibaba Cloud (Qwen) |
| Modality | Text in, text out only (no image or video input) |
| Context window | 1,000,000 tokens |
| Max output | 65,536 tokens |
| Reasoning | Extended-thinking (chain-of-thought) |
| Weights | Proprietary, closed-weight (not open source) |
| Status | Preview — announced May 20, 2026 |
| API compatibility | OpenAI-compatible; Anthropic-API compatible (works with Claude Code) |
| Model string | qwen3.7-max |
| GPTProto endpoint | https://gptproto.com/v1/chat/completions |
Qwen3.7-max vs GPT-5.5, Opus 4.8 & Qwen Turbo
All four run on GPTProto under one key and one balance, so you can route per task. GPT-5.5 and Opus 4.8 still lead the Artificial Analysis Intelligence Index; qwen3.7-max is the #1 Chinese model, trailing them narrowly while undercutting both heavily on price. Against Qwen Turbo — same 1M context, same text-only modality — qwen3.7-max trades speed for far deeper extended-thinking reasoning.
| Model | Modality | Context | Direct list (in/out per 1M) | On GPTProto |
|---|---|---|---|---|
| Qwen3.7-max | Text only | 1M | $2.50 / $7.50 | $0.36 / $1.44 |
| GPT-5.5 | Multimodal | ~1.05M | $5 / $30 | $4 / $24 |
| Claude Opus 4.8 | Multimodal | 1.0M | $5 / $25 | $4 / $20 |
| Qwen Turbo | Text only | 1M | $0.05 / $0.20 | $0.045 / $0.18 |
Honest take: Reach for qwen3.7-max when you want frontier-class text reasoning, a 1M context, and long agent runs at a fraction of the cost — on GPTProto roughly 80% below its official rate. For image, diagram, or video input, use a multimodal model like GPT-5.5, Opus 4.8, or qwen3.7-max's own sibling Qwen3.7-Plus.
Switching to the qwen3.7-max API on GPTProto
Moving from Alibaba Cloud Model Studio or OpenRouter to GPTProto is a drop-in change: keep your existing OpenAI-compatible request and swap two things — point base_url to https://gptproto.com/v1 and use your GPTProto API key. The same key and balance then cover qwen3.7-max and every other model on the platform, with no separate regional sign-up. Because Qwen3.7-max is also Anthropic-API compatible, Claude Code users can target it the same way.
