SOURCE-REVIEWED MODEL PROFILE
Qwen4-Max
2.4T-parameter sparse MoE flagship (~71B active per token) with 1M-token multimodal context and default reasoning, available via QwenCloud API.

Metadata reviewed 2026-11-16 · Original creator: Alibaba Qwen
PARAMETERS2.4T total (~71B active per token)
CONTEXT1,000,000 tokens
LICENSEProprietary (Cloud API)
MODALITYMultimodal to text
What it does
- 1M-token native context
- Text, image, and video inputs
- Adjustable reasoning effort
Local fit
Cloud API only (QwenCloud / compatible endpoints). No open weights or local checkpoint released for Qwen4-Max.
Supports text, image, and video inputs on QwenCloud.
Evidence and limits
Source reviewed; not locally tested.
License and source
Read the original model card ↗
Reviewed source revision: 2026-11-16