Qwen3.8-2.4T-A95BOpen Weight

Released

Qwen3.8-2.4T-A95B is an AI model released by Qwen on Wednesday, Aug 12 2026, 9 days after Qwen3.8-Max. It is an open-weight model — the trained weights are available to download and run. It is a 2.4T parameterThe learned values that shape how a model responds. B means billion and T means trillion. For mixture-of-experts models, the total count is larger than the number active for each token. Parameter count alone does not measure quality. model with a 262k token context windowHow much information the model can work with in one request, measured in tokens — pieces of words or other data. This includes your question, attachments, conversation and reply. Limits can vary by model variant and provider.. Benchmark results (shown below) cover Terminal-Bench 2.1, GPQA Diamond, and IFBench.

Benchmarks

Terminal & CLI

Terminal-Bench 2.1Agentic terminal coding — Can the AI work in a command-line terminal — running commands and finishing technical setup tasks the way a developer would? Higher is better.via NVIDIA BF16 evaluation
76.03%
#26 of 44

Reasoning & science

GPQA DiamondScience — Graduate-level science questions in biology, physics, and chemistry — hard enough that subject-matter PhDs score around 65%. Higher is better.via NVIDIA BF16 evaluation
92.55%
#16 of 145
IFBenchInstruction following — Tests whether the AI can follow detailed instructions and satisfy multiple constraints at once. Higher is better.via NVIDIA BF16 evaluation
79.93%
#4 of 16

Source: NVIDIA BF16 evaluation, retrieved 18 November 2026. Other sources are identified on the linked benchmark pages.

Compare Qwen3.8-2.4T-A95B with

Qwen3.8-2.4T-A95B

About

Qwen3.8-2.4T-A95B was released on August 12, 2026 as the first downloadable Qwen Max-class model. It used a 2.4-trillion-parameter sparse architecture with 95 billion active parameters per token and shipped under the Qwen3.8-Max License.

The public weights provided text-only reasoning and a native 262,144-token context, with a documented extension to roughly one million tokens. NVIDIA measured 92.55 on GPQA Diamond and 76.03 on Terminal-Bench 2.1 with the original BF16 weights.

Frequently asked questions

Qwen3.8-2.4T-A95B was released by Qwen on Wednesday, Aug 12 2026.

All Qwen releases

42 tracked