Yes, just — 8.6 GB spare

Llama 4 Scout 109B-A17B at Q4_K_M fits your M2 Max · 96 GB entirely in unified memory at 8K context, at an estimated 11 tokens per second. Past 191K the KV cache pushes it over — quantise the cache to q8_0, or step down a quantisation, to go longer.

Fully in unified memory 8K context Q4_K_M · 61.3 GB Llama 4 Community Released 5 Apr 2025 Vision

A 10M-token window on paper, chunked attention in practice (8K chunks on 3 of 4 layers). Needs 64 GB+ at Q4.

Hardware requirements What hardware do I need to run Llama 4 Scout 109B-A17B? →

At Q4_K_M with an 8K context: a 80 GB GPU (H100 SXM) or a 128 GB Mac. Minimum, recommended, high-quality and long-context tiers, with all 114 devices checked at every quantisation.

The VRAM budget

weights 61.3 GB
Weights 61.3 GB KV cache @ 8K 1.50 GB Runtime overhead 0.6 GB Free 8.6 GB of 72.0 GB

Quantisation ladder

QuantWeightsTotal @ 8KMax contextTok/sQualityFit
Q8_0 107.9 GB 110.0 GB — ~6.2 −0.1% ppl 38.0 GB over
Q6_K 83.2 GB 85.3 GB — ~8.0 −0.4% ppl 13.3 GB over
Q5_K_M 71.9 GB 74.0 GB — ~9.3 −0.8% ppl 2.0 GB over
Q4_K_M 61.3 GB 63.4 GB 191K 11 −1.9% ppl Recommended
Q3_K_M 49.6 GB 51.7 GB 440K 13 −5.4% ppl Long context
Q2_K 42.5 GB 44.6 GB 592K 16 −15% ppl Long context

Quality is the published perplexity delta against f16 weights. Max context assumes an f16 KV cache; q8_0 roughly doubles it. This model interleaves sliding-window layers (8192 tokens, 12 of 48 layers global), which is why its cache barely grows with context.

How to run it

terminal
$ pip install mlx-lm
$ mlx_lm.generate --model mlx-community/Llama-4-Scout-17B-16E-Instruct-4bit \
    --max-tokens 512 --prompt "Hello"

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

01Download is 61.3 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 72.0 GB of 96 GB. Raising it with iogpu.wired_limit_mb is possible, and risky.
03Only 8.6 GB is spare, so a long system prompt can still push it over. Its real ceiling here is 191K context.
See all models for this rig Compare with another model

Related models on an M2 Max · 96 GB

ModelTotalTok/s est.
Llama 3.3 70B Instruct
70.6B · same family
Q4_K_M · 42.8 GB 5.6 Runs great
Llama 3.1 8B Instruct
8.03B · same family
Q4_K_M · 6.1 GB 50 Runs great
Llama 3.2 3B Instruct
3.21B · same family
Q4_K_M · 3.3 GB 124 Runs great
Llama 3.2 1B Instruct
1.24B · same family
Q4_K_M · 1.5 GB 321 Runs great
gpt-oss 120B
117B MoE · similar size
MXFP4 · 61.4 GB 39 Tight fit
Mistral Small 4 119B-A6B
119B MoE · similar size
Q4_K_M · 67.7 GB 29 Tight fit
Nemotron 3 Super 120B-A12B
124B MoE · similar size
Q4_K_M · 70.4 GB 15 Tight fit

Every model against the M2 Max · 96 GB →

Llama 4 Scout 109B-A17B on other hardware

DeviceTotalTok/s est.
GeForce RTX 4060 Laptop
7.0 GB usable · 256 GB/s
63.4 GB ~1.6 CPU offload
GeForce RTX 3060 12 GB
10.6 GB usable · 360 GB/s
63.4 GB ~1.7 CPU offload
M4 · 16 GB
10.7 GB usable · 120 GB/s
63.4 GB — Won't fit
GeForce RTX 4090
22.4 GB usable · 1008 GB/s
63.4 GB ~2.1 CPU offload
Radeon RX 7900 XTX
22.4 GB usable · 960 GB/s
63.4 GB ~2.1 CPU offload
M3 Max · 64 GB
48.0 GB usable · 400 GB/s
63.4 GB — Won't fit
H100 SXM
78.4 GB usable · 3350 GB/s
63.4 GB 82 Runs great
DGX Spark · 128 GB
96.0 GB usable · 273 GB/s
63.4 GB 7.4 Runs great

All 114 devices, by tier →