Liquid AILFM3Lfm3OmniForConditionalGeneration

LFM3 Omni 2B — Hardware Requirements & GPU Compatibility

VisionChatFunctions

LFM3-Omni-2B is Liquid AI's first omni-modal model in the LFM3 family: a single end-to-end model that understands text, images, video, and audio and answers in text, speech, or both interleaved. It has about 2.2 billion parameters, built from the LFM3 backbone with an LFM3 vision encoder, a causal audio encoder, and an audio tokenizer. One model can stand in for separate speech recognition, vision-language, text, and speech synthesis models, and it can watch and listen to a video with sound at the same time. Liquid AI lists a 32K-token context for omni sessions. It is released under the LFM Open License v1.0, a custom license whose terms are in the repository, and was published in October 2026. It is designed to run efficiently on phones, laptops, and other edge hardware, so a single consumer GPU is more than enough.

2.2B params · dense128K contextOtherOct 20263.1K downloads227 likes47.8K quant downloads

Your hardware

Verdicts below are for the reference RTX 4090. Set yours to personalise the page.

Detection reads your browser's GPU name with no data sent anywhere.

Runs great on a 24 GB GPU

Runs from 8 GB (Q4_K_M · 125K)Best on 24 GB (Q4_K_M · ~385 tok/s · 125K)Macs from 8 GB (Q4_K_M · ~42 tok/s)

Other

Reference GPU · RTX 4090 24 GB

Estimates, ±20% · how we estimate

Run LFM3 Omni 2B

One recommended way. Every other method, OS and quant is in the step-by-step guide.Each way opens a step-by-step guide for your OS and quant.

Ways to runOther ways

each opens step-by-step

Run in cloud

Runs on RTX 3060 12GB (9 GB headroom) · Q4_K_M

Generation speed: ~138 tok/s; generation speed
Cost per 1M output tokens: $0.12; per 1M output tokens

Deployon Vast.aiBest valuefrom $0.061/hr

Compare GPUs →

Run LFM3 Omni 2B

2.4 GB VRAM at 16K context · fits the RTX 4090 (21.6 GB free)

Works on macOS, Windows and Linux.

Install Ollama (Linux)
$curl -fsSL https://ollama.com/install.sh | sh
Install Ollama (macOS / Windows)
Download and run the app; it installs the ollama command.
Download Ollama ↗
Run Q4_K_M
$ollama run hf.co/LiquidAI/LFM3-Omni-2B-GGUF:Q4_K_M
Raise the context window
Type it inside the chat. For the API server instead, start it with OLLAMA_CONTEXT_LENGTH=16384 ollama serve.
$/set parameter num_ctx 16384
A larger context needs more memory than the weights alone.
Use the API
Ollama serves its API on http://localhost:11434.

Works on macOS, Windows and Linux.

Install LM Studio
Desktop app for macOS (Apple silicon), Windows and Linux.
lmstudio.ai ↗
Download the model
In the app press Ctrl/Cmd + Shift + M and paste https://huggingface.co/LiquidAI/LFM3-Omni-2B-GGUF, then pick Q4_K_M. Or use the CLI (run LM Studio once first):
$lms get https://huggingface.co/LiquidAI/LFM3-Omni-2B-GGUF@Q4_K_M
Open on Hugging Face ↗
Load and chat
Open the model loader, select the model and set Context Length to 16384. Leave GPU offload at maximum if it fits.
Start the local server (optional)
$lms server start

Works on macOS, Windows and Linux.

Install llama.cpp (macOS)
MacPorts, Nix and conda-forge also work.
$brew install llama.cpp
Install options ↗
Install llama.cpp (Windows)
$winget install llama.cpp
Install llama.cpp (Linux)
Download the prebuilt llama-*-bin-ubuntu-* archive for your GPU backend (CPU, Vulkan) from the releases page, unpack it and run the binaries from that folder. Homebrew (brew install llama.cpp) and conda-forge also work on Linux.
llama.cpp releases ↗
Start the server
Downloads the model on first run. The chat UI and an OpenAI-compatible API are at http://localhost:8080.
$llama-server -hf LiquidAI/LFM3-Omni-2B-GGUF:Q4_K_M -c 16384 -ngl 99 --port 8080

Works on macOS, Windows and Linux.

Option A: Docker Model Runner
Docker Desktop: Settings > AI > Enable Docker Model Runner. Docker Engine on Linux: install the docker-model-plugin package.
$docker model run hf.co/LiquidAI/LFM3-Omni-2B-GGUF:Q4_K_M
Model Runner docs ↗
The quant tag after the repo is how Hugging Face cards write it; if your Docker version rejects it, drop the tag and pick the quant by repo.
Option B: llama.cpp server image (NVIDIA)
Needs the NVIDIA Container Toolkit. Models are cached in the llama-cache volume. Open http://localhost:8080.
$docker run --gpus all -p 8080:8080 -v llama-cache:/root/.cache ghcr.io/ggml-org/llama.cpp:server-cuda -hf LiquidAI/LFM3-Omni-2B-GGUF:Q4_K_M -c 16384 -ngl 99 --host 0.0.0.0 --port 8080
Image tags (cuda13, vulkan, rocm) ↗
Cache path inside the image is not in the docs; remove the -v flag if the volume complains.

Linux only.

Install vLLM
Linux only; on Windows run it inside WSL2. NVIDIA GPU shown; AMD ROCm, Intel and CPU builds have their own install pages.
$uv venv --python 3.12 --seed
$source .venv/bin/activate
$uv pip install vllm --torch-backend=auto
vLLM installation ↗
Serve the model
OpenAI-compatible API at http://localhost:8000/v1.
$vllm serve LiquidAI/LFM3-Omni-2B --max-model-len 16384
Runs the original weights, not the Q4_K_M GGUF; needs ~5 GB of GPU memory.

macOS only.

MLX supports only macOS, so follow these steps for that setup.

Install mlx-lm
$pip install mlx-lm
Chat in the terminal
$mlx_lm.chat --model LiquidAI/LFM3-Omni-2B-MLX-4bit
Downloads the model on first run. The MLX repo has its own quantisation; it may not match the quant selected above.
Or start a local server
Listens on http://localhost:8080.
$mlx_lm.server --model LiquidAI/LFM3-Omni-2B-MLX-4bit

Works on macOS, Windows and Linux.

Install Jan
jan.ai ↗
Find the model
Open the Hub tab, paste this model ID in Search, and choose Q4_K_M if offered.
$LiquidAI/LFM3-Omni-2B-GGUF
Use the local API (optional)
Settings > Local API Server > Start Server. It listens on http://127.0.0.1:1337.

Rent GPUs hourly to run Q4_K_M right there. Pricing comes from partner marketplaces and varies.

Available GPU rentals for Q4_K_M, prioritized by value
GPU	Speed	$/1M tokens	Rent
RTX 3060 12GBCheapestCheapest9 GB free	~138 tok/s	$0.12	Deploy on Vast.aifrom $0.061/hr (opens in new tab)	9 GB free · ~138 tok/s · $0.12 / 1M tokens
RTX 3060 Ti (8 GB)5 GB free	~171 tok/s	$0.12	Deploy on Vast.aifrom $0.072/hr (opens in new tab)	8 GB · 5 GB free · ~171 tok/s · $0.12 / 1M tokens
RTX 4060 (8 GB)5 GB free	~104 tok/s	$0.21	Deploy on Vast.aifrom $0.080/hr (opens in new tab)	8 GB · 5 GB free · ~104 tok/s · $0.21 / 1M tokens
RTX A4000 (16 GB)13 GB free	~171 tok/s	$0.15	Deploy on Vast.aifrom $0.092/hr (opens in new tab)	16 GB · 13 GB free · ~171 tok/s · $0.15 / 1M tokens
RTX 3070 (8 GB)5 GB free	~171 tok/s	$0.21	Deploy on RunPodfrom $0.13/hr (opens in new tab)	8 GB · 5 GB free · ~171 tok/s · $0.21 / 1M tokens
RTX 3090 (24 GB)21 GB free	~358 tok/s	$0.17	Deploy on RunPodfrom $0.22/hr (opens in new tab)	24 GB · 21 GB free · ~358 tok/s · $0.17 / 1M tokens
RTX A6000 (48 GB)45 GB free	~294 tok/s	$0.31	Deploy on RunPodfrom $0.33/hr (opens in new tab)	48 GB · 45 GB free · ~294 tok/s · $0.31 / 1M tokens
RTX 4090 (24 GB)21 GB free	~385 tok/s	$0.25	Deploy on RunPodfrom $0.34/hr (opens in new tab)	24 GB · 21 GB free · ~385 tok/s · $0.25 / 1M tokens

Compare GPUs →

How Much VRAM Does LFM3 Omni 2B Need?

Select a quantization level; the listed hardware adheres to it.

VRAM	+32K	+128K	File	Quality	On RTX 4090 (reference)
1.2 GB	2.8 GB	—	0.8 GB	3-bit · i-quant — Importance-weighted 3-bit	Fits: 125K context · ~528 tok/s
1.3 GB	2.8 GB	—	~0.9 GB	2-bit · K-quant — 2-bit quantization with K-quant improvements	Fits: 125K context · ~496 tok/s
1.3 GB	2.8 GB	—	0.9 GB	3-bit · i-quant small — Importance-weighted 3-bit, small	Fits: 125K context · ~496 tok/s
1.5 GB	3 GB	—	0.9 GB	3-bit medium — 3-bit medium quantization	Fits: 125K context · ~449 tok/s
1.5 GB	3 GB	—	1 GB	4-bit · legacy — 4-bit legacy quantization	Fits: 125K context · ~443 tok/s
1.5 GB	3 GB	—	1 GB	3-bit large — 3-bit large quantization	Fits: 125K context · ~434 tok/s
1.6 GB	3.1 GB	—	1.1 GB	4-bit · i-quant compact — Importance-weighted 4-bit, compact	Fits: 125K context · ~417 tok/s
1.6 GB	3.1 GB	—	1.1 GB	4-bit small — 4-bit small quantization	Fits: 125K context · ~404 tok/s
1.7 GB	3.2 GB	—	1.1 GB	4-bit medium — 4-bit medium quantization — most popular sweet spot	Fits: 125K context · ~385 tok/s
1.9 GB	3.4 GB	—	1.2 GB	5-bit small — 5-bit small quantization	Fits: 125K context · ~347 tok/s
1.9 GB	3.5 GB	—	1.3 GB	5-bit medium — 5-bit medium quantization — good quality/size tradeoff	Fits: 125K context · ~338 tok/s
2.2 GB	3.7 GB	—	1.6 GB	6-bit, very good — 6-bit quantization, very good quality	Fits: 125K context · ~299 tok/s
2.6 GB	4.1 GB	—	1.8 GB	8-bit, near-lossless — 8-bit quantization, near-lossless	Fits: 125K context · ~255 tok/s
4.7 GB	6.2 GB	—	3.5 GB	16-bit · bfloat — Brain floating point 16 — preferred for training	Fits: 125K context · ~139 tok/s

est.calculated, no published file yet

Selected Q4_K_M

4-bit medium quantization

2.4GBat 16K context

Weights: 1.3 GB
KV cache: 0.8 GB
Overhead: 0.3 GB

Fits RTX 4090 (reference) · 21.6 GB free · ~385 tok/s

≥50 tok/s needs ≥131 GB/s · RTX 5090 ~685, RTX 3090 Ti ~385

Which Hardware Can Run LFM3 Omni 2B?

Q4_K_M · 1.7 GB

LFM3 Omni 2B (Q4_K_M) requires 1.7 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 3+ GB is recommended. Using the full 128K context window can add up to 6.2 GB, bringing total usage to 7.9 GB. 52 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti. 163 devices with unified memory can run LFM3 Omni 2B, including Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB).

By VRAM tier · typical GPUs and Macs at Q4_K_M

Typical hardwareQuantVRAMtok/sContext that fits

8 GB · RTX 4060Q4_K_M · 2.4 GB · ~104 tok/s · 125K2.4~104125K8 GB tier →

12 GB · RTX 3060 12GBQ4_K_M · 2.4 GB · ~138 tok/s · 125K2.4~138125K12 GB tier →

16 GB · RTX 5060 Ti 16GBQ4_K_M · 2.4 GB · ~171 tok/s · 125K2.4~171125K16 GB tier →

24 GB · RTX 4090referenceQ4_K_M · 2.4 GB · ~385 tok/s · 125K2.4~385125K24 GB tier →

Mac 32 GB · Mac Mini M4Q4_K_M · 2.4 GB · ~49 tok/s · 125K2.4~49125KMacs →

Mac 64 GB · Mac Studio M4 MaxQ4_K_M · 2.4 GB · ~225 tok/s · 125K2.4~225125KMacs →

Alternatives to LFM3 Omni 2B

LFM3 family · 3 models →

Similar models

6tok/s on RTX 4090

Model	Score	VRAM	tok/s	Why it is similar
Qwen3.5 2BAlibaba	—	1.9 GB	~370	same size
Qwen3 VL 2B InstructAlibaba	—	3.5 GB	~362	same size
SmolVLM2 2.2B InstructHugging Face	—	2.3 GB	~443	same size
SmolVLM InstructHugging Face	—	4.9 GB	~320	same size
Qwen2 VL 2B InstructAlibaba	—	2.1 GB	~390	same size
InternVL2 2BOpenGVLab	—	3.3 GB	~449	same size

Specifications

Publisher: Liquid AI
Family: LFM3
Parameters: 2.2B
Architecture: Lfm3OmniForConditionalGeneration
Context Length: 128,000 tokens
Vocabulary Size: 128,000
Release Date: 2026-10-21
License: Other

Resources

Frequently Asked Questions

How much VRAM does LFM3 Omni 2B need?

LFM3 Omni 2B requires 1.7 GB of VRAM at Q4_K_M, or 4.7 GB at BF16. Full 128K context adds up to 6.2 GB (7.9 GB total).

VRAM = Weights + KV Cache + Overhead

Weights = 2.2B × 4.8 bits ÷ 8 = 1.3 GB

KV Cache + Overhead ≈ 0.4 GB (at 2K context + ~0.3 GB framework)

Fit ratings and hardware model lists check this model with room for a 16K-token context, which needs a little more memory.

KV Cache + Overhead ≈ 6.6 GB (at full 128K context)

VRAM usage by quantization

Q4_K_M

1.7 GB

Q4_K_M + full context

7.9 GB

Learn more about VRAM estimation →

What's the best quantization for LFM3 Omni 2B?

For LFM3 Omni 2B, Q4_K_M (1.7 GB) offers the best balance of quality and VRAM usage. Q5_K_S (1.9 GB) provides better quality if you have the VRAM. The smallest option is IQ3_XXS at 1.2 GB.

VRAM requirement by quantization

IQ3_XXS

1.2 GB~70%

Q3_K_M

1.5 GB~83%

Q4_K_S

1.6 GB~88%

Q4_K_M ★

1.7 GB~89%

Q5_K_M

1.9 GB~92%

BF16

4.7 GB~100%

★ Recommended — best balance of quality and VRAM usage.

Learn more about quantization →

Can I run LFM3 Omni 2B on a Mac?

Yes — MacBook Air 13" M3 (8 GB) and 145 other Macs can run LFM3 Omni 2B. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 1.7 GB of usable unified memory (RAM minus macOS overhead).

Can I run LFM3 Omni 2B locally?

Yes — LFM3 Omni 2B can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.

How fast is LFM3 Omni 2B?

At Q4_K_M, LFM3 Omni 2B can reach ~2824 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~385 tok/s. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.

tok/s = (bandwidth GB/s ÷ model GB) × efficiency

Example: NVIDIA B200 → 8000 ÷ 1.7 × 0.65 = ~3059 tok/s

Estimated speed at Q4_K_M (1.7 GB)

NVIDIA B200

~3059 tok/s

NVIDIA GeForce RTX 4090

~385 tok/s

NVIDIA B300

~3059 tok/s

AMD Instinct MI350X

~2824 tok/s

Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.

Learn more about tok/s estimation →

What's the download size of LFM3 Omni 2B?

At Q4_K_M, the download is about 1.30 GB. The full-precision BF16 version is 4.33 GB. The smallest option (IQ3_XXS) is 0.84 GB.

Which GPUs can run LFM3 Omni 2B?

52 consumer GPUs can run LFM3 Omni 2B at Q4_K_M. Top options include AMD Radeon RX 6700 XT, AMD Radeon RX 6800, AMD Radeon RX 6800 XT. 52 GPUs have plenty of headroom for comfortable inference.

Which devices can run LFM3 Omni 2B?

163 devices with unified memory can run LFM3 Omni 2B at Q4_K_M (1.7 GB), including AMD Ryzen AI 9 HX 370 (Strix Point) Laptop, ASUS Ascent GX10, Beelink GTR9 Pro (Ryzen AI Max+ 395, 128 GB), Framework Desktop (Ryzen AI Max+ 395, 128 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.