Qwen4-Max: capabilities and API setup
Explore Qwen4-Max for coding, document analysis, and multi-step tasks. Find official specifications, the exact Model ID, and BetterToken API settings.
Updated: November 17, 2026
Qwen4-Max is Qwen’s flagship multimodal model, released on November 16, 2026. Consider it for coding, frontend design, document analysis, and multi-step tasks. Its long context can accommodate extensive reference material; image and video understanding produces text, not generated images or video.
Official specifications
The following specifications describe Alibaba Cloud’s model. Use the BetterToken connection settings in the next section.
| Property | Official specification |
|---|---|
| Model ID | qwen4-max |
| Input / output | Text, images, video / text |
| Context | 1,000,000 tokens |
| Maximum input | 991,808 tokens without thinking; 983,616 with thinking |
| Maximum output | 131,072 tokens without thinking; 262,144 with thinking |
| Reasoning | Thinking and non-thinking modes; effort levels low, high (default), xhigh |
| Tools and structured output | Function calling, parallel tool calling, JSON Object, JSON Schema |
| Caching | Implicit and explicit caching |
Connect through BetterToken
Use your existing BetterToken API Key, or create one. Set Model ID to qwen4-max.
Use these APIs for non-streaming or streaming text conversations and tool calling:
| API | Base URL | Authentication header |
|---|---|---|
| Chat Completions | https://bettertoken.ai/v1 | Authorization: Bearer YOUR_BETTERTOKEN_API_KEY |
| Responses | https://bettertoken.ai/v1 | Authorization: Bearer YOUR_BETTERTOKEN_API_KEY |
| Anthropic Messages | https://bettertoken.ai | x-api-key: YOUR_BETTERTOKEN_API_KEY |
Chat Completions uses POST /v1/chat/completions, Responses uses POST /v1/responses, and Anthropic Messages uses POST /v1/messages.
Current pricing
Check the model plaza for current availability and input, output, and cache-read prices.

