Alibaba Cloud2.4T/95B active~1450 GB VRAM at Q4_K_M

Qwen3.8-Max — VRAM, speed & local setup

Written by Jakub Rusinowski · Last updated September 6, 2026

2.4 trillion parameters with roughly 95B active per token. Shipped as a service on 3 August 2026 with open weights following on 12 August under a model-specific Qwen3.8-Max licence, not Apache. Cluster-class self-hosting — about 1.45 TB at Q4_K_M — so for almost everyone this is an API model; it is listed here so the hardware checker can say so rather than stay silent.

Qwen3.8-Max needs about 1450 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.

Logic96

Creative93

Coding95

VRAM and speed by quantization

Quoted against NVIDIA RTX 4090 (24 GB). Weights plus overhead only: this model's architecture is unpublished, so no KV cache is added.

Quant	VRAM	Speed (est.)	Fit
Q2_K 2.63 bpw	789.8 GB	—	Too big
Q3_K_M 3.41 bpw	1023.8 GB	—	Too big
Q4_K_M 4.83 bpw	1449.8 GB	—	Too big
Q5_K_M 5.67 bpw	1701.8 GB	—	Too big
Q6_K 6.56 bpw	1968.8 GB	—	Too big
Q8_0 8.50 bpw	2550.8 GB	—	Too big
F16 16.00 bpw	4800.8 GB	—	Too big

Black marker = usable memory on the NVIDIA RTX 4090 (24 GB). Estimates from the memory-bandwidth roofline on the methodology page. Qwen3.8-Max VRAM calculator →

Get Qwen3.8-Max running

Qwen3.8-Max is not practical on a single consumer GPU: it needs about 1449.8 GB at Q4_K_M, more than any single device in our catalogue holds with headroom. How multi-GPU inference works, or rent a multi-GPU server by the hour:

Deploy in the cloud now RunPod

or compare on Vast.ai

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.

How to run Qwen3.8-Max

Install Ollama, then run:

$ollama run qwen3-8

Weights on Hugging Face: Qwen/Qwen3.8-2.4T-A95B ↗

Specifications

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

Parameters: 2.4 Trillion (95B active)
Context window: 262,144
Architecture: Mixture-of-Experts
Provider: Alibaba Cloud
Licence: Qwen3.8-Max License (model-specific)
Practical deployment: Private cloud / datacenter cluster
Specified at: Q4_K_M
System RAM: 2048 GB
Record updated: 2026-09-06

LicenceCustom Open-WeightCommercial use permitted

Weights are downloadable and commercial use is permitted, subject to the licence’s acceptable-use terms.

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 forreasoningenterprisecloud apiagentic tasks

Can I run Qwen3.8-Max on my GPU?

Other Qwen3.8 sizes

Qwen3.8-Max — frequently asked questions

How much VRAM does Qwen3.8-Max need?

About 1450 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 Qwen3.8-Max run on an RTX 4090 (24 GB)?

No. Qwen3.8-Max needs about 1450 GB at Q4_K_M, more than a single RTX 4090's 24 GB. It needs a larger card, several GPUs, or Apple Silicon with enough unified memory — or it runs with part of the weights offloaded to system RAM, which is much slower.

How do I run Qwen3.8-Max locally?

Install Ollama and run `ollama run qwen3-8`. That pulls the weights and starts a local OpenAI-compatible endpoint; after the download nothing leaves the machine.

What other sizes does Qwen3.8 come in?

Qwen3.8 27B (18 GB), Qwen3.8-Max (1450 GB), Qwen3.8-Flash-Next (109 GB). Every size shares the family's training and licence; the larger ones score higher and need proportionally more memory.