Qwen3.7 Max Explained: Price, Context and Specs
In short: Qwen3.7 Max is an AI model from Alibaba, released on 20 May 2026 and described by Alibaba as "the largest and most capable" model in the Qwen3.7 series. On Alibaba Cloud Model Studio's international region it costs $2.50 per million input tokens and $7.50 per million output. It has a 1,000,000-token context window, up to 131,072 tokens of output, and Alibaba's model page lists text input and output. Its successor is Qwen3.8-Max.
What is Qwen3.7 Max?
Qwen3.7 Max is the flagship of Alibaba's Qwen3.7 family, available through Alibaba Cloud Model Studio. Alibaba's model page calls Qwen3.7 "a next-generation flagship model designed for the agent-centric era", with strengths in programming, office and productivity tasks, and long-term autonomous execution. The API model ID is qwen3.7-max, which Alibaba says is functionally equivalent to the snapshot qwen3.7-max-2026-05-20.
For how models like this work, see how LLMs work.
How much does Qwen3.7 Max cost?
The price depends on the Model Studio region. From Alibaba's model page:
| Per 1M tokens | International (Singapore) | China (Beijing) |
|---|---|---|
| Input | $2.50 | $1.65 |
| Input (implicit cache) | $0.50 | $0.33 |
| Explicit cache creation | $3.125 | $2.063 |
| Explicit cache read | $0.25 | $0.165 |
| Output | $7.50 | $4.951 |
Alibaba's page shows original prices and notes that limited-time promotions are listed in the Model Studio console. Batch inference is listed as supported in the China (Beijing) region - at $0.825 input and $2.475 output per million tokens - and unsupported in the international and global regions. The full context window is billed at the same rate.
What are Qwen3.7 Max's specs?
| Spec | Qwen3.7 Max |
|---|---|
| Developer | Alibaba (Qwen team) |
| Released | May 20, 2026 |
| Model ID | qwen3.7-max (snapshot qwen3.7-max-2026-05-20) |
| Context window | 1,000,000 tokens |
| Maximum input | 991,808 tokens (983,616 in thinking mode) |
| Maximum output | 131,072 tokens |
| Maximum chain of thought | 262,144 tokens |
| Input / output | Text in, text out |
| Features | Function calling, structured outputs, web search, prefix completion, context caching |
| Fine-tuning | Not supported |
| Licence | Proprietary - not downloadable |
How does Qwen3.7 Max compare with Qwen3.8-Max?
Qwen3.8-Max is the later model, and its weights are published; Qwen3.7 Max is API-only.
| Qwen3.7 Max | Qwen3.8-Max | |
|---|---|---|
| Input / 1M tokens | $2.50 | $2 |
| Output / 1M tokens | $7.50 | $6 |
| Context window | 1M | 1M (hosted) |
| Input types | Text | Text, images, video (hosted) |
| Open weights | No | Yes, as Qwen3.8-2.4T-A95B |
How does Qwen3.7 Max rank?
On LMArena's text leaderboard Qwen3.7 Max has 1474 Elo (rank 42). Qwen3.8-Max has 1482 Elo (rank 24, max effort). The full ranked table is on our benchmarks page.
Words to know
- Snapshot - a fixed, dated version of a model that does not change.
- Implicit cache - automatic reuse of repeated input, billed at a lower rate.
- Explicit cache - input you ask the service to store for reuse, with a creation and a read price.
- Chain of thought - the reasoning a model writes before its final answer.
- Prefix completion - asking the model to continue from text you supply.
- Region - the data-centre location a cloud service runs in; prices can differ by region.
Related guides
Frequently asked questions
What is Qwen3.7 Max?+
Qwen3.7 Max is Alibaba's largest model in the Qwen3.7 series, released on 20 May 2026, designed for agent tasks such as programming, office work and long autonomous runs. It is available through Alibaba Cloud Model Studio.
How much does Qwen3.7 Max cost?+
In Model Studio's international region it costs $2.50 per million input tokens and $7.50 per million output tokens. In the China (Beijing) region it costs $1.65 and $4.951.
What is Qwen3.7 Max's context window?+
Qwen3.7 Max has a 1,000,000-token context window, with up to 991,808 input tokens and 131,072 output tokens per request.
Does Qwen3.7 Max support images?+
Alibaba's current model page lists text as the only input and output modality for qwen3.7-max, in every region.
Can you download Qwen3.7 Max?+
No. Qwen3.7 Max is available only as a hosted API service. Its successor's weights, Qwen3.8-2.4T-A95B, are published on Hugging Face.
Enjoyed this? Get more in your inbox.
Up to ten of the week's AI news stories, by email once a week.