Qwen3.8-2.4T-A95B
Qwen3.8-2.4T-A95B is Qwen's downloadable 2.4-trillion-parameter sparse mixture-of-experts model with 95B activated parameters.
Provider
Alibaba Qwen
Model family
Alibaba Qwen
Open-weight sparse MoE reasoning LLM
Cost tier
Open Flagship
Status
Current
Release Aug 12, 2026
Why teams choose it
The checkpoint activates about 95B of 2.4T total parameters per token and is a distinct
product from hosted Qwen3.8-Max.
The released checkpoint is text-only and always uses thinking
vision and non-thinking support belong to the hosted Max service.
Native context is 262,144 tokens
with documented extension to roughly 1.01M under supported serving configurations.
Tradeoffs to know
- The 2.4T checkpoint has exceptional storage, memory, networking, and serving requirements despite sparse activation.
- Its custom Qwen3.8-Max License is not Apache 2.0; review attribution and commercial-service obligations.
- Model-card capabilities and evaluations are vendor-reported and sensitive to serving stack, precision, and reasoning configuration.
Technical specs
- Inputs
- text
- Outputs
- text
- Capabilities
- reasoning, coding, research, long-horizon agents, long-context, configurable reasoning, local deployment
- License
- Qwen3.8-Max License
- Source
- https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B
- Native context
- 262,144
- Total parameters
- 2.4T
- Activated parameters
- 95B
Benchmarks
No independent or vendor benchmark scores in this catalog entry.
Alibaba Qwen family lineup
Current models
Related models
Qwen3.8-2.4T-A95B FAQ
What is Qwen3.8-2.4T-A95B?
Qwen3.8-2.4T-A95B is Qwen's downloadable 2.4-trillion-parameter sparse mixture-of-experts model with 95B activated parameters. The official model card documents text-only input and output, mandatory thinking, 262,144 native context extensible to roughly 1.01M, configurable reasoning effort, and a dedicated Qwen3.8-Max license.
When does Qwen3.8-2.4T-A95B fit best?
Large-scale self-hosted reasoning
What should teams watch out for with Qwen3.8-2.4T-A95B?
The 2.4T checkpoint has exceptional storage, memory, networking, and serving requirements despite sparse activation.
Explore next
Models, tools, and comparisons that connect to this reference.