How much memory does Qwen3.5-397B-A17B need?

About 229 GB at 4-bit, 430 GB at 8-bit and 809 GB at full 16-bit precision, including an 8,192-token KV cache and about 1.5 GB of runtime overhead. These are estimates from its parameter count.

Can Qwen3.5-397B-A17B run on a 24 GB graphics card?

Not entirely: even at 4-bit it needs about 229 GB, so part of it would sit in system memory, which is much slower.

What is Qwen3.5-397B-A17B’s context window?

Its configuration sets 262,144 tokens. Longer contexts need more memory for the KV cache, so the practical limit on a given machine can be lower.

ANVE: the private AI office, on the record. See what that means

ANVE

Run it on your own hardware · updated 2026-10-16

Qwen3.5-397B-A17B
The memory it needs, what you may do with it, and the hardware it suits.

Short answer. Qwen3.5-397B-A17B needs about 229 GB of memory at 4-bit, 430 GB at 8-bit and 809 GB at full 16-bit precision, including runtime overhead and an 8,192-token KV cache. It has about 403.4 billion parameters and a 262,144-token context window. Licence in short: commercial use allowed. By memory alone, on ANVE’s launch configurations: ANVE One does not fit, ANVE Office does not fit, ANVE Office Pro (GPU) does not fit. These are estimates from its published size, not benchmarks.

What it is

Qwen3.5-397B-A17B is a vision-language model: it reads images (scanned pages, photos, charts and screenshots) as well as text, published by Alibaba’s Qwen team on Hugging Face on 16 February 2026. It has about 403.4 billion parameters and a configured context window of 262,144 tokens. It is a mixture-of-experts model: 512 experts, 10 used for each token, so only about 17B parameters are active at a time.

Its card lists the licence as Apache 2.0. In short, the licence allows commercial use, modification and redistribution, with attribution and a copy of the licence, and includes a patent grant.

Hugging Face counted about 478,911 downloads of it in the 30 days before we checked (7 Oct 2026), and 1,564 people have liked it there.

How much memory it needs

We size it the way you would size any model: the weights (parameters × bits ÷ 8), the KV cache for 8,192 tokens of context, and roughly 1.5 GB for the runtime itself. Real use varies with settings.

Each token of context adds about 0.03 MB of KV cache at 16-bit, so 32,768 tokens would add about 1.0 GB. Only 15 of its 60 layers use standard attention; the others keep a small fixed state, so the cache is smaller than in a classic transformer.

Precision	Weights	KV cache (8,192 tokens)	Total, estimated
BF16 (full precision)	807 GB	0.25 GB	809 GB
8-bit (Q8_0)	429 GB	0.25 GB	430 GB
4-bit (Q4_K)	227 GB	0.25 GB	229 GB

Architecture at a glance

Setting	Value in config.json
Architecture class	Qwen3_5MoeForConditionalGeneration
Model type	qwen3_5_moe
Layers	60
Hidden size	4,096
Attention heads	32
Key-value heads	2 (grouped-query attention)
Head size	256
Vocabulary	248,320
Experts	512, 10 active per token
Configured context	262,144
Stored precision	bfloat16

Will it fit on ANVE?

These verdicts are arithmetic, not tests, against the specification at launch: ANVE One can give a model about three quarters of its 64 GB of unified memory, while ANVE Office and ANVE Office Pro hold it on a 24 GB graphics card and keep the rest of the stack in 64 GB or 128 GB of system memory. ANVE’s tested default is Qwen 3.8 27B; this model is not yet validated on ANVE.

Configuration (at launch)	Memory for the model	Verdict
ANVE One	about 48 GB of 64 GB unified memory	does not fit
ANVE Office	24 GB on the RTX PRO 4000 Blackwell SFF Edition	does not fit
ANVE Office Pro (GPU)	24 GB on the RTX PRO 4000 Blackwell	does not fit

What an office would use it for

This is frontier-scale: even at 4-bit it needs far more memory than any single office box, so it belongs on a multi-GPU server or a cloud service, not on an appliance.

Similar models to compare

MiniMax-M3 (427.0B parameters, about 242 GB at 4-bit before context; custom licence: check before commercial use)
MiniMax-VL-01 (456.4B parameters, about 258 GB at 4-bit before context; licence not stated)
GLM-5.3-Flash (321.3B parameters, about 182 GB at 4-bit before context; commercial use allowed)

Questions people ask

How much memory does Qwen3.5-397B-A17B need?: About 229 GB at 4-bit, 430 GB at 8-bit and 809 GB at full 16-bit precision, including an 8,192-token KV cache and about 1.5 GB of runtime overhead. These are estimates from its parameter count.
Can Qwen3.5-397B-A17B run on a 24 GB graphics card?: Not entirely: even at 4-bit it needs about 229 GB, so part of it would sit in system memory, which is much slower.
Is Qwen3.5-397B-A17B free for commercial use?: Its card lists Apache 2.0, which allows commercial use, modification and redistribution, with attribution and a copy of the licence, and includes a patent grant. Read the licence itself before relying on this summary; it is not legal advice.
What is Qwen3.5-397B-A17B’s context window?: Its configuration sets 262,144 tokens. Longer contexts need more memory for the KV cache, so the practical limit on a given machine can be lower.

What is not real yet

No sealed ANVE unit exists; every image of the appliance on this site is a labelled render. The software runs today on a development machine. No customers yet, no certifications, no delivery date, and the final price is not announced.

Reservations are ₹999 in India or US$12 elsewhere, refundable in full until you confirm an order in writing (terms). Offices that cannot use cloud AI can apply to be one of ten design partners.

Qwen3.5-397B-A17BThe memory it needs, what you may do with it, and the hardware it suits.