Skip to content
Weights Not Walls
Menu

Qwen

Qwen3.8-2.4T-A95B

Reviewed
Size
2.4T parameters, 95.0B active (from the repo name)counted from the safetensors header on the Hub
Licence
custom, read it, Qwen3.8-Max LicenseOur reading, check the licence: A custom licence we have not read for this model: other makers use the same name for different texts. Read the model's own licence file before you use it.Licence text
On the Hub since
8 Aug 2026the date the repo was created
Context length
256k tokensfrom the model's config
Kind
Base model (no base model declared on the Hub)Qwen3_5MoeForCausalLM
On Hugging Face
Qwen/Qwen3.8-2.4T-A95B30k downloads in the last 30 days, 1.4k likes

Scores

No independent scores for this model in Epoch AI's data. A score is only shown here when Epoch's name for a model matches this repo with certainty; an uncertain match is left out rather than guessed.

Claimed by the maker Claimed

The model card has no claimed scores in a machine-readable form (the card's model-index), so none are shown. The maker may give scores in the card text: read the card.

Other views, which we link to and do not copy: look this model up on Artificial Analysis and the Arena leaderboard.

Run it yourself

file sizes read from the Hub 19 Nov 2026

Memory needed = weights + context cache + 0.50 GB for the runtime. Context cache at 8k tokens: 0.72 GB, worked out from the model's config (Qwen/Qwen3.8-2.4T-A95B), at 16-bit precision. Choose a machine above to see a verdict for each format. Method

GGUF for llama.cpp, Ollama, LM Studio and others

unsloth/Qwen3.8-2.4T-A95B-GGUF (converted by unsloth), 11k downloads. Sizes are the real files on the Hub, all parts added up.

FormatSizeMemory neededGet it
UD-Q1_0370.0 GBfile size, 10 parts371.2 GB370.0 + 0.72 + 0.50
Commands
llama.cpp
llama-server -hf unsloth/Qwen3.8-2.4T-A95B-GGUF:UD-Q1_0
Download
hf download unsloth/Qwen3.8-2.4T-A95B-GGUF --include "UD-Q1_0/Qwen3.8-2.4T-A95B-UD-Q1_0-*-of-00010.gguf" --local-dir Qwen3.8-2.4T-A95B-GGUF
UD-IQ1_S473.5 GBfile size, 12 parts474.7 GB473.5 + 0.72 + 0.50
Commands
llama.cpp
llama-server -hf unsloth/Qwen3.8-2.4T-A95B-GGUF:UD-IQ1_S
Download
hf download unsloth/Qwen3.8-2.4T-A95B-GGUF --include "UD-IQ1_S/Qwen3.8-2.4T-A95B-UD-IQ1_S-*-of-00012.gguf" --local-dir Qwen3.8-2.4T-A95B-GGUF
UD-IQ1_M525.2 GBfile size, 13 parts526.4 GB525.2 + 0.72 + 0.50
Commands
llama.cpp
llama-server -hf unsloth/Qwen3.8-2.4T-A95B-GGUF:UD-IQ1_M
Download
hf download unsloth/Qwen3.8-2.4T-A95B-GGUF --include "UD-IQ1_M/Qwen3.8-2.4T-A95B-UD-IQ1_M-*-of-00013.gguf" --local-dir Qwen3.8-2.4T-A95B-GGUF
UD-IQ2_XXS611.5 GBfile size, 15 parts612.7 GB611.5 + 0.72 + 0.50
Commands
llama.cpp
llama-server -hf unsloth/Qwen3.8-2.4T-A95B-GGUF:UD-IQ2_XXS
Download
hf download unsloth/Qwen3.8-2.4T-A95B-GGUF --include "UD-IQ2_XXS/Qwen3.8-2.4T-A95B-UD-IQ2_XXS-*-of-00015.gguf" --local-dir Qwen3.8-2.4T-A95B-GGUF
UD-IQ2_XS680.5 GBfile size, 16 parts681.7 GB680.5 + 0.72 + 0.50
Commands
llama.cpp
llama-server -hf unsloth/Qwen3.8-2.4T-A95B-GGUF:UD-IQ2_XS
Download
hf download unsloth/Qwen3.8-2.4T-A95B-GGUF --include "UD-IQ2_XS/Qwen3.8-2.4T-A95B-UD-IQ2_XS-*-of-00016.gguf" --local-dir Qwen3.8-2.4T-A95B-GGUF
UD-IQ3_XXS889.9 GBfile size, 21 parts891.1 GB889.9 + 0.72 + 0.50
Commands
llama.cpp
llama-server -hf unsloth/Qwen3.8-2.4T-A95B-GGUF:UD-IQ3_XXS
Download
hf download unsloth/Qwen3.8-2.4T-A95B-GGUF --include "UD-IQ3_XXS/Qwen3.8-2.4T-A95B-UD-IQ3_XXS-*-of-00021.gguf" --local-dir Qwen3.8-2.4T-A95B-GGUF
UD-IQ4_XS1220.8 GBfile size, 29 parts1222.1 GB1220.8 + 0.72 + 0.50
Commands
llama.cpp
llama-server -hf unsloth/Qwen3.8-2.4T-A95B-GGUF:UD-IQ4_XS
Download
hf download unsloth/Qwen3.8-2.4T-A95B-GGUF --include "UD-IQ4_XS/Qwen3.8-2.4T-A95B-UD-IQ4_XS-*-of-00029.gguf" --local-dir Qwen3.8-2.4T-A95B-GGUF
Q8_02421.7 GBfile size, 56 parts2422.9 GB2421.7 + 0.72 + 0.50
Commands
llama.cpp
llama-server -hf unsloth/Qwen3.8-2.4T-A95B-GGUF:Q8_0
Download
hf download unsloth/Qwen3.8-2.4T-A95B-GGUF --include "Q8_0/Qwen3.8-2.4T-A95B-Q8_0-*-of-00056.gguf" --local-dir Qwen3.8-2.4T-A95B-GGUF
BF164557.1 GBfile size, 140 parts4558.3 GB4557.1 + 0.72 + 0.50
Commands
llama.cpp
llama-server -hf unsloth/Qwen3.8-2.4T-A95B-GGUF:BF16
Download
hf download unsloth/Qwen3.8-2.4T-A95B-GGUF --include "BF16/Qwen3.8-2.4T-A95B-BF16-*-of-00140.gguf" --local-dir Qwen3.8-2.4T-A95B-GGUF

MLX for Apple silicon

FormatSizeMemory neededRepo
MLX 4bitpipenetwork/Qwen3.8-2.4T-A95B-MLX-reap75-4bit336.1 GBfile size337.4 GB336.1 + 0.72 + 0.50Download page
MLX 3bitkernelpool/Qwen3.8-2.4T-A95B-3bit-UVMAX750.3 GBfile size751.5 GB750.3 + 0.72 + 0.50Download page

Original weights size from the safetensors header, for vLLM, SGLang, Transformers

FormatSizeMemory neededRepo
BF16original weights4556.4 GBfile size4557.6 GB4556.4 + 0.72 + 0.50
hf download Qwen/Qwen3.8-2.4T-A95B --local-dir Qwen3.8-2.4T-A95B
Other conversions (11; sizes not read)

The Hub tags 34 quantised repos as conversions of this model; we read the most downloaded and show the maker's own first.

Run it by API

Hugging Face Inference Providers, fetched 19 Nov 2026, 13:58 UTC

4 providers serve this model through the Hugging Face router. Prices are US dollars per million tokens. Precision is what the provider reports; a lower precision than the original weights can change answers. Live first, then cheapest output.

ProviderInOutContextPrecisionTokens/sFirst tokenUptime 24hToolsJSON
deepinfra$2.00$6.00256knot given690.7 s?yesyes
Novita$2.00$6.00977knot given451.3 s99.7%yesyes
SiliconFlow$2.00$6.001Mfp8331.4 s99.5%yesyes
Venice$2.00$6.00256knot given1232.5 s99.7%yesno

Through the router, with a Hugging Face token (the router bills your Hugging Face account). Change the name after the colon to pick another provider.

curl https://router.huggingface.co/v1/chat/completions \ -H "Authorization: Bearer $HF_TOKEN" -H "Content-Type: application/json" \ -d '{"model": "Qwen/Qwen3.8-2.4T-A95B:deepinfra", "messages": [{"role": "user", "content": "Hello"}]}'

Not built yet: one key for every open model, with BlockRun, and our own tested-quality score per provider. Until then, precision and measured speed are the best guide to a provider here.

Planned for this page Not built yet

Try it in private chat
A link to Solana AI's confidential chat for the models it serves. This model is not listed as served there.
Fine-tune it
Fine-tuning a few base models from this page and publishing the result here.
Creators and earnings
Pages for people who publish fine-tunes and adapters of this model, and what they earn.
Fund and bounties
Bounties for new open models, paid from a public fund. How the fund will work.
Contribute compute
How to give compute to open model work, with Akash.

Model data from the Hugging Face Hub 19 Nov 2026, 13:43 UTC. The repo. Found a mistake on this page? Tell us.