Model library / Qwen
Qwen3.8-2.4T-A95B is a 2.4 trillion-parameter causal language model with 95 billion active parameters, supporting up to 262 k token context.
Mixture of experts, count not published
Context length in tokens, as published
Weights at the released precision
Licence declared on the repository
Overview
Qwen3.8-2.4T-A95B is a causal language model from Qwen, built on the Qwen3.5 architecture and post-trained for coding, professional work, research and long-horizon agentic tasks. The publisher highlights improved autonomous planning, flexible reasoning control and broad compatibility with development tools and ease of integration into existing pipelines.
The model contains 2.4 trillion parameters overall, with 95 billion activated per inference, organised in 92 layers and a hidden dimension of 8192. It uses a mixture of 512 experts, routing ten plus one shared per token, and runs in bfloat16 format. Native context length is 262 k tokens, extendable to about one million, and it is released under a non-standard licence.
Specification
| Repository | Qwen/Qwen3.8-2.4T-A95B |
| Publisher | Qwen |
| Published | 8 August 2026 |
| Architecture | Qwen3_5MoeForCausalLM |
| Model type | qwen3_5_moe_text |
| Parameters | 2,446,182,725,504 |
| Layers | 92 |
| Hidden size | 8,192 |
| Attention heads | 64, 4 key/value heads |
| Context length | 262,144 tokens |
| Vocabulary | 248,320 tokens |
| Weight format | bfloat16 |
| Licence | other |
Memory
Parameter count multiplied by the bytes each format uses. The key-value cache comes on top.
| Precision | Weights | Notes |
|---|---|---|
| bf16 | 4556.4 GiB | Full precision as released |
| fp8 | 2278.2 GiB | 8-bit, near-lossless on most models |
| int4 | 1139.1 GiB | 4-bit, smallest footprint |
The cache costs 368 KiB per token at 16-bit, so the full 262,144-token context of one request needs about 92.0 GiB. Concurrency multiplies that number, not the weights.
Hardware
At the released precision, with 90% of the card given to the server and 2.5 GiB kept for the runtime. The context column is what is left for the cache on that setup.
| GPU | Cards needed | Free for cache | Context it holds |
|---|---|---|---|
| RTX 4090 24GB | more than 8 | — | — |
| RTX 5090 32GB | more than 8 | — | — |
| L40S 48GB | more than 8 | — | — |
| A100 80GB | more than 8 | — | — |
| H100 80GB | more than 8 | — | — |
| RTX PRO 6000 96GB | more than 8 | — | — |
| H200 141GB | more than 8 | — | — |
Serving
A vLLM launch line for the shape above. Check the model card for a runtime the publisher recommends.
vllm serve Qwen/Qwen3.8-2.4T-A95B \ --tensor-parallel-size 8 \ --max-model-len 262144 \ --gpu-memory-utilization 0.90
At the released precision the weights do not fit on eight of the largest cards above, so this needs more than one node, or a quantised build.
Questions
The weights take about 4556 GiB at the released precision. No single card in the table above holds that, so it needs several GPUs or a lower precision.
The repository declares other. Read the licence text before commercial use: the name alone does not tell you what is allowed.
The configuration allows 262,144 tokens. Whether the whole window is usable depends on the memory left for the cache, which the hardware table works out per card.
This page is a reference, and Aiqre does not serve this model today. We deploy open-weight models on dedicated EU hardware on request, so if you want an OpenAI-compatible endpoint for it, email [email protected] with the model name and roughly what volume you expect.
Also in the library
Nex-N2.5-Max is a 1.6-trillion-parameter text-only MoE model designed for long-horizon agentic tasks.
DeepSeek-V4-Pro-0813 is a 1650.5B mixture-of-experts open-weight model published by deepseek-ai on Hugging Face. It uses the Deep…
Hosting
Tell us the model and roughly what volume you expect. We reply with a price and an OpenAI-compatible endpoint, with your prompts never stored.