LLMs · Liquid AI

LFM2 24B-A2B

Name: LFM2 24B-A2B
Author: Liquid AI

Parameters: 24B (2.3B active)
Architecture: Mixture of experts
Max context: 125K tokens
License: lfm1.0
Released: 2026-02

How to run it

Quickest · Ollama

One command downloads a ready-to-run version and starts a chat.

ollama run lfm2:24b

Get Ollama at ollama.com/download (opens in a new tab) · model page (opens in a new tab)

Quantized files · GGUF

Works in LM Studio, llama.cpp, Jan and Ollama. No sign-in needed.

huggingface.co/LiquidAI/LFM2-24B-A2B-GGUF (opens in a new tab)

ollama run hf.co/LiquidAI/LFM2-24B-A2B-GGUF

Original weights · Hugging Face

Full-precision files from Liquid AI, for fine-tuning, vLLM, Transformers.

huggingface.co/LiquidAI/LFM2-24B-A2B (opens in a new tab)

No sign-in needed.

Official repo · GitHub

Reference code, inference scripts and docs from Liquid AI. The weights themselves download from Hugging Face.

github.com/Liquid4All/cookbook (opens in a new tab)

git clone https://github.com/Liquid4All/cookbook

Liquid AI’s official cookbook for running LFM models.

Links checked 6 days ago against Hugging Face, Ollama and GitHub.

Memory needed

Weights plus KV cache and runtime overhead. Leave a little extra free for the operating system and other apps.

KV cache: hybrid convolution + attention: only the attention layers keep a KV cache.

Quantization	Quality	Weights	Total at 8K context	Total at 32K context	Download
`Q2_K`	Low	9.5 GB	10.3 GB	10.8 GB	GGUF files (opens in a new tab)
`Q3_K_M`	Low	11.7 GB	12.4 GB	13.0 GB	GGUF files (opens in a new tab)
`Q4_K_M`	Good	14.5 GB	15.3 GB	15.8 GB	GGUF (opens in a new tab)
`Q5_K_M`	High	16.9 GB	17.7 GB	18.2 GB	GGUF (opens in a new tab)
`Q6_K`	High	19.5 GB	20.3 GB	20.8 GB	GGUF (opens in a new tab)
`Q8_0`	Very high	25.3 GB	26.0 GB	26.5 GB	GGUF (opens in a new tab)
`F16`	Full	47.7 GB	48.4 GB	49.0 GB	GGUF (opens in a new tab) Original weights (opens in a new tab)

Best hardware for LFM2 24B-A2B

The least expensive GPU in our catalog that runs it without offloading is the Intel Arc B580 (launch price $249) at Q2_K.

Grade · points	Hardware	Quant	Memory	Speed	Fit
79 ptsGreat (A) · 79 of 100 points: Model quality 24.9/45, Memory fit 25.0/25, Speed 15.0/15, Memory use 15.0/15	NVIDIA GeForce RTX 5090	`Q6_K`	21.9 GB / 32.0 GB	332 tok/s	Runs well
79 ptsGreat (A) · 79 of 100 points: Model quality 24.9/45, Memory fit 25.0/25, Speed 15.0/15, Memory use 14.9/15	Apple M4 Pro (48 GB)	`Q6_K`	23.2 GB / 36.0 GB	43.5 tok/s	Runs well
79 ptsGreat (A) · 79 of 100 points: Model quality 25.2/45, Memory fit 25.0/25, Speed 15.0/15, Memory use 14.4/15	Apple M1 Max (64 GB)	`Q8_0`	29.9 GB / 48.0 GB	46.8 tok/s	Runs well
79 ptsGreat (A) · 79 of 100 points: Model quality 25.2/45, Memory fit 25.0/25, Speed 15.0/15, Memory use 14.4/15	Apple M4 Max (64 GB)	`Q8_0`	29.9 GB / 48.0 GB	63.9 tok/s	Runs well
79 ptsGreat (A) · 79 of 100 points: Model quality 24.4/45, Memory fit 25.0/25, Speed 15.0/15, Memory use 15.0/15	AMD Radeon RX 7900 XTX	`Q5_K_M`	18.9 GB / 24.0 GB	182 tok/s	Runs well
79 ptsGreat (A) · 79 of 100 points: Model quality 24.4/45, Memory fit 25.0/25, Speed 15.0/15, Memory use 15.0/15	NVIDIA GeForce RTX 3090	`Q5_K_M`	18.9 GB / 24.0 GB	206 tok/s	Runs well
79 ptsGreat (A) · 79 of 100 points: Model quality 24.4/45, Memory fit 25.0/25, Speed 15.0/15, Memory use 15.0/15	NVIDIA GeForce RTX 4090	`Q5_K_M`	18.9 GB / 24.0 GB	222 tok/s	Runs well
77 ptsGreat (A) · 77 of 100 points: Model quality 25.2/45, Memory fit 25.0/25, Speed 15.0/15, Memory use 12.8/15	NVIDIA RTX PRO 6000 Blackwell	`F16`	53.2 GB / 96.0 GB	112 tok/s	Runs well
76 ptsGreat (A) · 76 of 100 points: Model quality 21.2/45, Memory fit 25.0/25, Speed 15.0/15, Memory use 15.0/15	AMD Radeon RX 9070 XT	`Q3_K_M`	13.2 GB / 16.0 GB	190 tok/s	Runs well
76 ptsGreat (A) · 76 of 100 points: Model quality 21.2/45, Memory fit 25.0/25, Speed 15.0/15, Memory use 15.0/15	NVIDIA GeForce RTX 4060 Ti 16GB	`Q3_K_M`	13.2 GB / 16.0 GB	99.4 tok/s	Runs well

Check it on your own hardware

Estimates come from published hardware specs and our open formulas, not from running the model on your machine. Real speed varies with runtime, drivers and settings. Numbers marked “Measured” come from a linked benchmark.