text model · LFM · macOS

Can I run LFM2.5 8B-A1B on Apple M4 Pro (24GB)?

Compatibility verdict VRAM threshold engine

Yes, it runs GPU accelerated

Yes. LFM2.5 8B-A1B runs on Apple M4 Pro (24GB) at Q4_K_M (~6.7 GB of ~16 GB usable).

Needs ~6.7 GB Device usable ~16 GB

Runs at Q4_K_M using ~6.7 GB of ~16 GB usable. You have room for Q8_0 for higher quality.

That figure is at a 4k context and moves about ±15% as context length changes. Apple M4 Pro (24GB) leaves ~9.3 GB of headroom, room to step up to Q8_0 for higher quality.

Q4_K_M needed: ~6.7 GB
Usable on device: ~16 GB
Device memory: 24 GB
Best quant: Q4_K_M

Share on X Share on Reddit

Which quant fits

Quant ladder vs ~16 GB usable

Q2_K

~5 GB

Q3_K_M

~5.6 GB

Q4_K_M

~6.7 GB

Q5_K_M

~7.4 GB

Q6_K

~8.3 GB

Q8_0

~10.5 GB

FP16

~18.5 GB

The line marks Apple M4 Pro (24GB)'s ~16 GB budget; rungs past it are too large.

Run it

Install commands macOS

Pick your tool. All 3 load the same Q4_K_M weights.

Ollama

$ ollama run lfm2.5:8b-a1b

llama.cpp

$ llama-cli -hf LiquidAI/LFM2.5-8B-A1B-GGUF:Q4_K_M

LM Studio

$ lms get LiquidAI/LFM2.5-8B-A1B-GGUF

Model LFM

Parameters: 8.3B (MoE, 1.5B active)
Q4_K_M size: 5.2 GB
Q8_0 size: 9 GB
Context: 128k
Ollama tag: lfm2.5:8b-a1b

Full LFM2.5 8B-A1B requirements →

Device macOS

Memory: 24 GB unified
Usable for weights: ~16 GB
Power draw: ~140 W
Best runtime: Ollama (MLX backend, preview) / MLX direct

Best models for Apple M4 Pro (24GB) →

You could also run

Run LFM2.5 8B-A1B on other hardware

Nvidia GeForce RTX 3060 (12GB)Nvidia GeForce RTX 4070 (12GB)Samsung Galaxy S24 Ultra Generic Android Phone (12GB RAM)iPhone 17 Pro iPhone Air

FAQ

Can Apple M4 Pro (24GB) run LFM2.5 8B-A1B?

Yes. LFM2.5 8B-A1B runs on Apple M4 Pro (24GB) at Q4_K_M (~6.7 GB of ~16 GB usable).

How much memory does LFM2.5 8B-A1B need?

Apple M4 Pro (24GB) has room to spare. At Q4_K_M the weights are ~5.2 GB; with KV cache and runtime overhead, budget ~6.7 GB at a 4k context. It is a Mixture-of-Experts model (8.3B total / 1.5B active), so all experts must stay in memory; memory tracks total params, not active params.

What is the best tool to run LFM2.5 8B-A1B on macOS?

LM Studio for a simple setup; mlx-lm for the most speed. vLLM is NOT a Mac tool, it is a CUDA/Linux serving engine. Unified memory is not a fixed VRAM slice; ~70% is usable for weights.

Embed this

A live badge for your model card or README, updated as the data is.

Markdown

[![LFM2.5 8B-A1B on Apple M4 Pro (24GB)](https://localmodel.run/badge/lfm2.5-8b-a1b/apple-m4-pro-24gb.svg)](https://localmodel.run/can-i-run/lfm2.5-8b-a1b/apple-m4-pro-24gb)

Sources

Weights are measured from GGUF files; KV cache and overhead are computed, so totals can vary ~15% with context and runtime. Any tok/s is a bandwidth estimate. See methodology.