Qwen3.7 max

qwen3.7-max

Qwen3.7-Max is the flagship model of Alibaba's Qwen3.7 series. It supports text input and output, is designed for agent-centric workloads, and offers unique advantages in coding, office and productivity tasks, and long-term autonomous execution. Compared with previous generations of Qwen, this model achieves significant improvements in coding and agent performance, and supports explicit prompt caching for efficient reuse of repeated contexts.

Context window1.0M

Provider通义千问

Released2026/06/29

Playground

Pricing

The same model is available through multiple service channels — choose based on latency, reliability and cost.

Prices in $ / 1M tokens

To pick a channel, add a provider field to the request body, for example "provider": { "channel": "direct" }. Valid values are direct / stable / economical; omit it to use the default channel.

Direct

Direct upstream connection — best when you need native behavior and the full context window.

Input	Output
1.76/M	5.29/M

Overview

Input

Text

Output

Text


# Qwen3.7-Max API: Alibaba Qwen's Trillion-Parameter Flagship Agent Model

Qwen3.7-Max is the flagship model in Alibaba's Qwen3.7 series, featuring a trillion-parameter MoE architecture, officially released at the Alibaba Cloud Summit in May 2026. It supports text input and output and is designed for agent-centric workloads, with distinctive advantages in coding, office and productivity tasks, and long-horizon autonomous execution.

Compared with previous generations of Qwen, Qwen3.7-Max has made significant progress in coding and agent performance, and supports explicit prompt caching for efficient reuse of repeated context. In the Arena AI comprehensive evaluation, Qwen3.7-Max-Preview scored 1284, surpassing GPT-5.5's 1215.

SeaWhale AI provides Qwen3.7-Max through an OpenAI-compatible interface, supporting tool calling, prompt caching, and streaming output.

> [Get API Key](https://api.atalk-ai.com) · Model ID: `qwen3.7-max`

---

## Why Choose Qwen3.7-Max

- **Arena AI overall score 1284** — surpassing GPT-5.5's 1215
- **Trillion-parameter MoE architecture** — large-scale sparse expert model with a high capability ceiling
- **Agent-oriented design** — optimized for long-horizon autonomous workloads
- **SWE-bench Verified 72.3%** — agent coding capability ranks first in China and third globally
- **1 million-token context** — fit large repositories and long document sets at once
- **Explicit prompt caching** — significantly lower cost for reusing repeated context

---

## Core Capabilities

### 01 Agentic Coding
Qwen3.7-Max achieves 72.3% on SWE-bench Verified, ranking first in China and third globally. This is a result on real software engineering tasks—understanding the repository, locating the problem, writing the fix, and passing verification.

- Repository-level problem localization and fixing
- Cross-file feature implementation
- Autonomous execution of long-running coding tasks

### 02 Long-Horizon Autonomous Execution
"Long-horizon autonomous execution" is a core design goal of Qwen3.7-Max. It can maintain goal consistency in long-chain tasks, repeatedly calling tools, checking results, and adjusting the path until delivery is complete.

- Goal maintenance in long-chain tasks
- Tool calling and result verification
- Path replanning after intermediate failures

### 03 Office and Productivity Tasks
Office scenarios such as document processing, spreadsheet generation, and report writing are one of Qwen3.7-Max's strengths, and also a natural advantage for domestic models in local business scenarios.

- Structured document and report generation
- Spreadsheet data processing and analysis
- Professional writing in Chinese contexts

### 04 Explicit Prompt Caching
Support for explicit prompt caching means that reused system prompts, knowledge bases, and code contexts are paid for at full price only once, with subsequent calls significantly reducing cost—this is especially important in agent scenarios, where a large amount of context must be resent every round.

- Low-cost reuse of repeated context
- Cost optimization for multi-round agent calls
- Cache hits for long system prompts

---

## Best Use Cases

| Scenario | Description |
|---|---|
| Agentic coding | Repository-level bug fixing and feature development |
| Long-horizon autonomous tasks | Automated workflows that need to maintain goals for extended periods |
| Office automation | Batch generation of documents, spreadsheets, and reports |
| Professional Chinese writing | High-quality content production in local contexts |
| Long-context analysis | Large-scale material processing within a 1 million-token window |
| Domestic substitution | Flagship choice under data compliance requirements |

---

## Differences Between Qwen3.7-Max, Qwen3.7-Plus, and Qwen3.6-Max

| Capability | **Qwen3.7-Max** | Qwen3.7-Plus | Qwen3.6-Max Preview |
|---|---|---|---|
| Model ID | **`qwen3.7-max`** | `qwen3.7-plus` | `qwen3.6-max-preview` |
| Architecture | **Trillion-parameter MoE** | Dense architecture | Hybrid architecture |
| Input modalities | **Text** | Text, image | Text, image |
| Context window | **1 million tokens** | 1 million tokens | 1 million tokens |
| Max output | **128K tokens** | 128K tokens | 65.5K tokens |
| Arena AI | **1284** | 1156 | — |
| Focus | **Agentic coding and autonomous execution** | Multimodal interactive agent | Frontend and repository-level issues |

> The specific billing is subject to the real-time price card at the top of the page.

---

## How to Use the Qwen3.7-Max API

**1. Create a SeaWhale AI API key**
Generate a key and top up your balance in the console.

**2. Make good use of prompt caching**
In agent scenarios, a large amount of context must be resent every round. Put stable, unchanged parts (system prompts, knowledge bases, code context) at the front and enable caching—the cost can drop by an order of magnitude.

**3. Call the API**

```bash
curl -X POST https://api.atalk-ai.com/v1/chat/completions \
  -H 'Authorization: Bearer YOUR_API_KEY' \
  -H 'Content-Type: application/json' \
  -d '{
    "model": "qwen3.7-max",
    "messages": [
      {"role": "user", "content": "Analyze the test coverage of this repository and add unit tests for critical paths."}
    ],
    "stream": true
  }'

FAQ

When was Qwen3.7 released? It was officially released at the Alibaba Cloud Summit on May 20, 2026, including Qwen3.7-Max-Preview (trillion-parameter MoE) and Qwen3.7-Plus-Preview (dense architecture).

What level is an Arena AI score of 1284? It surpasses GPT-5.5's 1215 and places it in the first tier in comprehensive capability evaluations.

How good is the coding capability? SWE-bench Verified 72.3%, first in China and third globally.

Does it support multimodal? Qwen3.7-Max is a text-only model (text input, text output). For image input, please use Qwen3.7-Plus.

What are the context and output limits? 1 million tokens of context, with a maximum output of 128K tokens.

How do I use prompt caching? Put stable, unchanged context at the front of the request. The model will explicitly cache this part, and subsequent calls with the same prefix will be billed at the cache price.

Why Choose SeaWhale AI for the Qwen3.7-Max API

No Alibaba Cloud account required — call directly with one API key, no need to separately activate Bailian
OpenAI-compatible interface — integrate with just two lines of changes to existing code
Cross-vendor comparison — compare performance and cost with GPT, Claude, and GLM under the same account
Settlement in RMB — domestic models are priced in RMB for clear accounting

API

API integration

Model IDUse this value as the model in inference requests

qwen3.7-max

API KeyBearer token used to authenticate inference requests

Manage API keys

Base URLOpenAI compatible · /chat/completions

https://api.atalk-ai.com/v2/

provider OptionalSelects a service channel; omit it and the system picks the default

"provider": { "channel": "direct" }

qwen3.7-max usage examples

SeaWhale AI is compatible with the OpenAI API protocol, so you can call it with the OpenAI SDK or plain HTTP requests. Streaming is enabled by default.

About the provider parameter (optional, a SeaWhale AI extension): most models are served over several channels that differ slightly in price and reliability. Add a provider field to the request body to pick one; omit it and the system selects the default channel — normal calls are unaffected.

provider is not part of the official OpenAI protocol — it is a SeaWhale AI extension that only takes effect on this platform. The OpenAI SDK allows custom fields like this to pass through; see the examples below.

Value	Channel	Best for
`direct`	Direct	The official upstream link, for native behavior and the full context window
`stable`	Preferred	Balanced availability and speed — a good default for production traffic
`economical`	Economy	Cost first, well suited to batch processing and price-sensitive workloads

Available channels and their prices are listed under "Pricing" above (channels vary by model). Additional notes:

Syntax: "provider": { "channel": "direct" }.
If the requested channel is not enabled for that model, the request automatically falls back to the default channel and returns normally — it does not error.
With the official SDKs: in Python pass it via extra_body; in Node.js put it directly on the request object and it passes through. In TypeScript projects, add a // @ts-expect-error line to skip the type check.

cURLPythonNode.js

curl https://api.atalk-ai.com/v2/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer <API_KEY>" \
  -d '{
    "model": "qwen3.7-max",
    "messages": [
      {"role": "system", "content": "You are a helpful assistant."},
      {"role": "user", "content": "Hello!"}
    ],
    "provider": { "channel": "direct" },
    "stream": true
  }'
# provider is optional — remove this line to use the default channel

from openai import OpenAI

client = OpenAI(
    base_url="https://api.atalk-ai.com/v2",
    api_key="<API_KEY>",
)

stream = client.chat.completions.create(
    model="qwen3.7-max",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Hello!"},
    ],
    stream=True,
    # Optional: pick a service channel; omit to use the default
    extra_body={"provider": {"channel": "direct"}},
)

for chunk in stream:
    if chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="", flush=True)

import OpenAI from 'openai'

const client = new OpenAI({
  baseURL: 'https://api.atalk-ai.com/v2',
  apiKey: '<API_KEY>',
})

const stream = await client.chat.completions.create({
  model: 'qwen3.7-max',
  messages: [
    { role: 'system', content: 'You are a helpful assistant.' },
    { role: 'user', content: 'Hello!' },
  ],
  stream: true,
  // Optional: pick a service channel; omit to use the default
  // @ts-expect-error provider is a SeaWhale AI extension, not in the OpenAI SDK types
  provider: { channel: 'direct' },
})

for await (const chunk of stream) {
  process.stdout.write(chunk.choices[0]?.delta?.content ?? '')
}