Alibaba Cloud35B (3B active)~22 GB VRAM at Q4_K_M

Qwen 3.6 35B-A3B — VRAM, speed & local setup

Name: LLM Configurator — GPU VRAM Checker
Author: LLM Configurator

Written by Jakub Rusinowski · Last updated April 16, 2026

MoE successor to Qwen 3.5 35B-A3B, now with native image AND video input (3.5 was image-only). Q4_K_M is ~21 GB — fits a 24GB GPU, though a 32GB card (RTX 5090) leaves more headroom for KV cache at long context. Community reports running it on 6GB VRAM via heavy CPU offload at ~30 tok/s.

Qwen 3.6 35B-A3B needs about 22 GB of VRAM at Q4_K_M — quantized weights plus framework overhead, before any KV cache. On Apple Silicon that figure comes out of unified memory.

Logic92

Creative90

Coding93

VRAM and speed by quantization

Quoted against NVIDIA RTX 4090 (24 GB). Includes the KV cache at 8K context, so it reads higher than the headline figure.

Quant	VRAM	Speed (est.)	Fit
Q2_K 2.63 bpw	12.5 GB	~267 tok/s	Fits
Q3_K_M 3.41 bpw	16 GB	~242 tok/s	Fits
Q4_K_M 4.83 bpw	22.2 GB	~207 tok/s	Tight
Q5_K_M 5.67 bpw	25.8 GB	~29 tok/s	Offload
Q6_K 6.56 bpw	29.7 GB	~26 tok/s	Offload
Q8_0 8.50 bpw	38.2 GB	~23 tok/s	Offload
F16 16.00 bpw	71 GB	—	Too big

Black marker = usable memory on the NVIDIA RTX 4090 (24 GB). Estimates from the memory-bandwidth roofline on the methodology page. Qwen 3.6 35B-A3B VRAM calculator →

Get Qwen 3.6 35B-A3B running

The cheapest catalogued GPU that runs Qwen 3.6 35B-A3B is the NVIDIA GeForce RTX 5090 (32 GB).

Buy this hardware NVIDIA GeForce RTX 5090 32GB — 32 GB VRAM · 575 W board power Deploy in the cloud now RTX 4090 on RunPod — from $0.34/hr · rate checked 2026-07

or compare on Vast.ai from $0.35/hr (typical low · varies)

As an Amazon Associate we earn from qualifying purchases. Cloud GPU links are referral links — we may earn a commission at no extra cost to you.

Affiliate disclosure: Some links on this page are affiliate links — if you buy through them, LLM Configurator may earn a commission at no extra cost to you. As an Amazon Associate, LLM Configurator earns from qualifying purchases.

NVIDIA GeForce RTX 5090 32GB

32 GB VRAM · 575 W board power

2026 prices are volatile — check the current listing.

Check price on Amazon

How to run Qwen 3.6 35B-A3B

Install Ollama, then run:

$ollama run qwen3.6:35b-a3b

Weights on Hugging Face: Qwen/Qwen3.6-35B-A3B ↗

Specifications

Verified — Checked against the primary source — the model card or the vendor spec page — and corroborated by a second independent source.

Parameters: 35 Billion (3B active)
Context window: 262,144
Architecture: Hybrid Gated DeltaNet + MoE (256 experts)
Provider: Alibaba Cloud
Licence: Apache 2.0
Practical deployment: Single consumer GPU
Specified at: Q4_K_M
System RAM: 32 GB
Record updated: 2026-04-16

LicenceApache-2.0Commercial use permitted

Commercial use permitted. No usage restrictions beyond attribution.

Quality and use cases

Scores as published by the model’s authors or an independent evaluator — quality, not throughput, and not measured by us.

Best forreasoningcodingmultimodalconsumer gpu

Can I run Qwen 3.6 35B-A3B on my GPU?

Other Qwen 3.6 sizes

Qwen 3.6 27B~18 GB at Q4_K_M

Qwen 3.6 35B-A3B — frequently asked questions

How much VRAM does Qwen 3.6 35B-A3B need?

About 22 GB at Q4_K_M — quantized weights plus framework overhead, before any KV cache. The cache grows with context length and is added on top; the table above folds it in. Apple Silicon counts unified memory toward the same figure.

Does Qwen 3.6 35B-A3B run on an RTX 4090 (24 GB)?

Yes. Qwen 3.6 35B-A3B needs about 22 GB at Q4_K_M, inside a 24 GB card, at an estimated 207 tokens/sec.

How do I run Qwen 3.6 35B-A3B locally?

Install Ollama and run `ollama run qwen3.6:35b-a3b`. That pulls the weights and starts a local OpenAI-compatible endpoint; after the download nothing leaves the machine.

What other sizes does Qwen 3.6 come in?

Qwen 3.6 27B (18 GB), Qwen 3.6 35B-A3B (22 GB). Every size shares the family's training and licence; the larger ones score higher and need proportionally more memory.