Qwen3.6 35B-A3B (MoE): VRAM requirements and which GPUs run it

Qwen3.6 35B-A3B (MoE) has 36B parameters with 3B active per token (mixture of experts). It has 40 layers, of which only 10 are full attention (2 KV heads of dimension 256) and keep a cache that grows with the conversation; the other 30 are Gated DeltaNet linear-attention layers with a small fixed state. At the everyday setting, Q4_K_M and 8k context, it needs an estimated 22.4 GB, so the smallest card that fits is 24 GB.

What Qwen3.6 35B-A3B (MoE) is

The first open-weight Qwen3.6 model, April 2026: 256 experts with 8 routed and one shared, 3B parameters read per token. The release notes put agentic coding and repository-level reasoning first.

Text, images and video 262,144-token window Apache 2.0 April 2026 3B of 36B active

Reasoning. Supports thinking, with an option to keep reasoning from earlier turns in context.

Memory by quantisation and context

Weights at this quant: 20.9 GB at Q4. Every extra 1,000 tokens of context adds about 0.02 GB of KV cache at FP16, on top of a fixed 0.06 GB of recurrent state that does not grow. Try other settings in the VRAM calculator.

Which GPU to pick for Qwen3.6 35B-A3B (MoE)

The smallest card here that loads it at Q4_K_M with 8k of context is the RTX 3090 (24 GB): 22.4 GB needed, an estimated 377 tokens per second, and a tight fit that leaves little for the conversation. For room to work, meaning 32,768 tokens of context while staying under 85% of memory, step up to the RTX 5090 (32 GB), which needs 22.9 GB for that and generates at about 721 tokens per second.

At Q8_0, which is near-lossless, the card is the L40S (48 GB), needing 40.4 GB. On Apple silicon, a MacBook Pro M4 Max holds it with 32,768 tokens of context at an estimated 220 tokens per second; prompt processing is slower than on Nvidia, which long documents make obvious. Among unified-memory desktops, the Ryzen AI Max+ 395 holds it at 32,768 tokens in 22.9 GB, at an estimated 103 tokens per second. To check a card that is not on this page, or a different context length, use the can-I-run-it checker, which opens on this model.

Speed by GPU, at Q4 and 8k context

Single-stream decode, from memory bandwidth at 70% efficiency. Prompt processing and batching not included. These are ceilings, not measurements: Qwen3.6 35B-A3B (MoE) reads only 1.7 GB of weights per token at Q4, so little that the per-token costs the formula leaves out decide much of the real speed. The 11 figures marked ≤ are where a real runtime falls furthest below the number shown. See the speed estimator for other quantisations and Apple hardware, or every card compared if you are choosing hardware rather than a model.

What Qwen3.6 35B-A3B (MoE) is good at, and what it is not

Good at

• Fast agentic coding where the latency of a dense model would hurt.
• A 32 GB card, where it has both the speed and the room for a long context.
• Vision-language work at mixture-of-experts speed.

Watch out for

• On a 24 GB card it is a tight fit: it loads, with little room left.
• All 36B of weights sit in memory to read 3B per token. It is fast, not small.
• The hybrid layers need a current runtime.

Similar models

The models that need about the same memory as Qwen3.6 35B-A3B (MoE) at Q4_K_M and 8k, which makes them the real alternatives on whatever card you have. The context column is what each extra 1,000 tokens costs, and it is where models of the same size differ most.

Notes

• All 256 experts stay in memory; 3B are active per token. The parameter count includes the vision encoder; a text-only GGUF is slightly smaller.
• Native context window: 262,144 tokens. Licence: Apache 2.0, a permissive licence.
• Weights published in April 2026. It follows Qwen3 30B-A3B (MoE).
• Architecture values from config.json in Qwen/Qwen3.6-35B-A3B on Hugging Face, and its model card. Every figure on this page is computed from those values; check them before buying hardware for this model.
• Estimates assume a single conversation on a card that is otherwise free. A desktop on the same GPU takes 0.5 to 2 GB.

Questions

Can I run Qwen3.6 35B-A3B (MoE) on a 24 GB card?

Yes, at Q4_K_M and 8k context it needs about 22.4 GB, which fits a 24 GB card tightly.

How much VRAM does Qwen3.6 35B-A3B (MoE) need at Q8?

About 40.4 GB at 8k context, or 40.9 GB at 32k. Q8 is near-lossless; use it when it fits.

What is the context length of Qwen3.6 35B-A3B (MoE)?

262,144 tokens natively. Holding all of it at Q4_K_M takes about 27.6 GB, of which 5.4 GB is context.

Why does Qwen3.6 35B-A3B (MoE) need so little memory for long context?

Because most of its layers do not keep a cache that grows. It has 40 layers, of which only 10 are full attention (2 KV heads of dimension 256) and keep a cache that grows with the conversation; the other 30 are Gated DeltaNet linear-attention layers with a small fixed state. At 32k tokens the context costs about 0.7 GB; if all 40 layers kept a conventional full-length cache, the same conversation would cost about 2.7 GB.

How fast is Qwen3.6 35B-A3B (MoE) on an RTX 4090?

At most about 406 tokens per second at Q4, single stream, which is faster than you can read. That is a ceiling worked out from the card's memory bandwidth, not a measurement, and at a figure this high real runtimes land well below it, because the costs the formula leaves out take a growing share of each token.

See every model compared and which models fit on 8 to 96 GB, or run it without buying the card: Nodegrove attaches a 24, 48 or 96 GB GPU to a workspace that stays saved.