Qwen3.8-Max Release: 2.4 Trillion Parameters and Open Weights
Qwen released Qwen3.8-Max on August 3, 2026. This text model features 2.4 trillion parameters and plans to release open-source weights next week.
Official Release Details
The official announcement occurred on August 3, 2026. On this date, Qwen stated that they were officially releasing Qwen3.8-Max. The provider characterized this specific release as the most capable model in the Qwen family to date. The source material confirms that the model operates with text input and text output modalities.
- Release date: August 3, 2026
- Model name: Qwen3.8-Max
- Provider: Qwen
- Input/Output: Text
Architecture and Parameter Scale
According to the release notes, Qwen3.8-Max is built upon the architectural foundation of Qwen 3.5. The documentation specifies that the new model scales to 2.4 trillion parameters. The source excerpt mentions comprehensive improvements across coding, work, and research, though it does not provide specific benchmark scores or quantitative performance metrics for these areas. The context window size and maximum output token limits are not specified in the provided facts.
- Base architecture: Qwen 3.5
- Parameter count: 2.4 trillion
- Claimed focus areas: Coding, work, research
Weight Availability and Evaluation Steps
The release marks the first time Qwen will open-source the weights of a Qwen-Max-class model. The source states that these open weights will be released the week following the initial launch. Currently, the model is listed with openWeights set to false, indicating that public weight files are not yet available for immediate download. Engineers can evaluate the model by accessing the service through the provider's interface. To validate performance, users should define specific test cases relevant to their workflow, measure the output quality against existing baselines, and compare results before the open weights become available. This process allows for a practical assessment of the model's capabilities without relying on unverified architectural claims.
- Open weights status: Planned for next week
- Current access: Via provider service
- Evaluation method: Test, measure, and compare outputs
Artificial Analysis benchmark review
Benchmark review updated on . These scores apply to the linked artifact and recorded methodology; they do not establish the performance of a different runtime or quantized model.
| Measurement | Recorded result | Measured |
|---|---|---|
| Artificial Analysis Coding Index (v4.3) ↗ | 76.2 index_points | 2026-09-16 |
| Artificial Analysis Intelligence Index (v4.3) ↗ | 45.4 index_points | 2026-09-16 |
Review the full benchmark provenance and source methodology →
Primary sources
Frequently asked questions
When will the open-source weights be available?
The source indicates that the open-source weights for Qwen3.8-Max are planned for release the week after the initial launch on August 3, 2026. As of the discovery date, the weights have not been confirmed as publicly downloadable.
What are the known specifications of Qwen3.8-Max?
The model has 2.4 trillion parameters and uses text for both input and output. It is built on the Qwen 3.5 architecture. Specific details regarding pricing, context window length, and maximum output tokens are not provided in the verified source facts.