No — not on this device

Qwen3.8 2.4T-A95B at Q4_K_M needs 1376.7 GB against 384.0 GB usable, and the shortfall of 992.7 GB is more than 128 GB of system RAM can cover at a tolerable speed. A smaller sibling or a lower quantisation is the honest answer here.

Does not fit 8K context Q4_K_M · 1375.4 GB Qwen3.8-Max License Released 12 Aug 2026 Not in the Ollama library

The first open Qwen-Max-class flagship. Listed as the honest ceiling: nothing short of a rack runs it.

Hardware requirements What hardware do I need to run Qwen3.8 2.4T-A95B? →

No single GPU or Mac in the catalogue holds it at Q4_K_M — the page says what comes closest. Minimum, recommended, high-quality and long-context tiers, with all 114 devices checked at every quantisation.

Fits instead: Qwen3.8-Flash-Next 180B-A6B (102.0 GB) · Qwen3.8 27B (16.7 GB)

The VRAM budget

weights 1375.4 GB
Weights 1375.4 GB KV cache @ 8K 0.72 GB Runtime overhead 0.6 GB Over budget 992.7 GB past 384.0 GB

Quantisation ladder

QuantWeightsTotal @ 8KMax contextTok/sQualityFit
Q8_0 2420.4 GB 2421.7 GB — ~2.2 −0.1% ppl 2037.7 GB over
Q6_K 1868.0 GB 1869.3 GB — ~2.9 −0.4% ppl 1485.3 GB over
Q5_K_M 1614.5 GB 1615.9 GB — ~3.3 −0.8% ppl 1231.9 GB over
Q4_K_M 1375.4 GB 1376.7 GB — ~3.9 −1.9% ppl 992.7 GB over
Q3_K_M 1113.4 GB 1114.7 GB — ~4.8 −5.4% ppl 730.7 GB over
Q2_K 953.9 GB 955.2 GB — ~5.6 −15% ppl 571.2 GB over

Quality is the published perplexity delta against f16 weights. Max context assumes an f16 KV cache; q8_0 roughly doubles it. Only 23 of its 92 blocks keep a per-token KV cache; the rest are linear-attention, Mamba or convolution blocks with a fixed-size state.

How to run it

terminal
$ pip install mlx-lm
$ mlx_lm.generate --model mlx-community/Qwen3.8-2.4T-A95B-4bit \
    --max-tokens 512 --prompt "Hello"

Apple's own array framework. The fastest path on Apple Silicon. More on MLX.

01Download is 1375.4 GB. Keep it on an SSD — a first load off a spinning disk takes minutes.
02macOS caps what the GPU may wire down at about 384.0 GB of 512 GB. Raising it with iogpu.wired_limit_mb is possible, and risky.
03If it falls back to CPU silently, drop the context first, then step down a quantisation.
See all models for this rig Compare with another model

Related models on an M3 Ultra · 512 GB

ModelTotalTok/s est.
Qwen3.8-Flash-Next 180B-A6B
180B MoE · same family
Q4_K_M · 102.0 GB 62 Runs great
Qwen3.8 27B
27.8B · same family
Q4_K_M · 16.7 GB 29 Runs great
Kimi K3 2.8T-A104B
2780B MoE · similar size
Q4_K_M · 1564.0 GB — Won't fit
DeepSeek V4 Pro 1.6T-A49B
1650B MoE · similar size
Q4_K_M · 928.9 GB — Won't fit
Kimi K2.6 1T-A32B
1027B MoE · similar size
Q4_K_M · 578.6 GB — Won't fit
GLM-5.2 744B-A40B
753B MoE · similar size
Q4_K_M · 424.7 GB — Won't fit
GLM-5.3 744B-A40B
753B MoE · similar size
Q4_K_M · 424.7 GB — Won't fit

Every model against the M3 Ultra · 512 GB →

Qwen3.8 2.4T-A95B on other hardware

DeviceTotalTok/s est.
GeForce RTX 4060 Laptop
7.0 GB usable · 256 GB/s
1376.7 GB — Won't fit
GeForce RTX 3060 12 GB
10.6 GB usable · 360 GB/s
1376.7 GB — Won't fit
M4 · 16 GB
10.7 GB usable · 120 GB/s
1376.7 GB — Won't fit
GeForce RTX 4090
22.4 GB usable · 1008 GB/s
1376.7 GB — Won't fit
Radeon RX 7900 XTX
22.4 GB usable · 960 GB/s
1376.7 GB — Won't fit
M3 Max · 64 GB
48.0 GB usable · 400 GB/s
1376.7 GB — Won't fit

All 114 devices, by tier →