Liquid AILFM2.5Lfm2ForCausalLM

LFM2.5 230M — Hardware Requirements & GPU Compatibility

ChatFunctions

LFM2.5-230M is Liquid AI's smallest model in the LFM2.5 family, a 230-million-parameter checkpoint distilled from the larger LFM2.5-350M and refined with multi-stage reinforcement learning for lightweight agentic tasks. It is instruction-tuned for data extraction and simple tool-use pipelines under the tightest memory and compute budgets, and the publisher does not recommend it for reasoning-heavy work like advanced math, code generation, or creative writing. It is released under Liquid AI's custom LFM1.0 license. At 230 million parameters, it runs on essentially any modern device, including phones and low-power boards like a Raspberry Pi, without needing a dedicated GPU at all.

230M params · dense128K contextOtherJun 202677.0K downloads311 likes439.8K 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 · ~1310 tok/s · 125K)Macs from 8 GB (Q4_K_M · ~143 tok/s)

Other

Reference GPU · RTX 4090 24 GB

Estimates, ±20% · how we estimate

Run LFM2.5 230M

One recommended way. Every other method, OS and quant is in the step-by-step guide.

Other ways

each opens step-by-step

Run in cloud

Fits on RTX 3060 12GB (11 GB headroom) · Q4_K_M

Generation speed: ~468 tok/s; generation speed
Cost per 1M output tokens: $0.045; per 1M output tokens

Deployon Vast.aiBest optionfrom $0.076/hr

Compare GPUs →

Run LFM2.5 230M

0.6 GB VRAM at 16K context · fits the RTX 4090 (23.4 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/LFM2.5-230M-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/LFM2.5-230M-GGUF, then pick Q4_K_M. Or use the CLI (run LM Studio once first):
$lms get https://huggingface.co/LiquidAI/LFM2.5-230M-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/LFM2.5-230M-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/LFM2.5-230M-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/LFM2.5-230M-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/LFM2.5-230M --max-model-len 16384
Runs the original weights, not the Q4_K_M GGUF; needs ~1 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/LFM2.5-230M-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/LFM2.5-230M-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/LFM2.5-230M-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 12GBCheapestCheapest11 GB free	~468 tok/s	$0.045	Deploy on Vast.aifrom $0.076/hr (opens in new tab)	11 GB free · ~468 tok/s · $0.045 / 1M tokens
Tesla V100 PCIe 16GB15 GB free	~1170 tok/s	$0.025	Deploy on Vast.aifrom $0.10/hr (opens in new tab)	15 GB free · ~1170 tok/s · $0.025 / 1M tokens
RTX A4000 (16 GB)15 GB free	~582 tok/s	$0.053	Deploy on Vast.aifrom $0.11/hr (opens in new tab)	16 GB · 15 GB free · ~582 tok/s · $0.053 / 1M tokens
RTX 3070 (8 GB)7 GB free	~582 tok/s	$0.062	Deploy on RunPodfrom $0.13/hr (opens in new tab)	8 GB · 7 GB free · ~582 tok/s · $0.062 / 1M tokens
RTX 4060 Ti 16GB15 GB free	~374 tok/s	$0.11	Deploy on Vast.aifrom $0.14/hr (opens in new tab)	15 GB free · ~374 tok/s · $0.11 / 1M tokens
RTX 3090 (24 GB)23 GB free	~1217 tok/s	$0.050	Deploy on RunPodfrom $0.22/hr (opens in new tab)	24 GB · 23 GB free · ~1217 tok/s · $0.050 / 1M tokens
RTX A6000 (48 GB)47 GB free	~998 tok/s	$0.092	Deploy on RunPodfrom $0.33/hr (opens in new tab)	48 GB · 47 GB free · ~998 tok/s · $0.092 / 1M tokens
RTX 4090 (24 GB)23 GB free	~1310 tok/s	$0.072	Deploy on RunPodfrom $0.34/hr (opens in new tab)	24 GB · 23 GB free · ~1310 tok/s · $0.072 / 1M tokens

Compare GPUs →

How Much VRAM Does LFM2.5 230M Need?

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

VRAM	+32K	+128K	File	Quality	On RTX 4090 (reference)
0.4 GB	0.8 GB	—	~0.1 GB	2-bit · i-quant extreme compression — Importance-weighted 2-bit, extreme compression — significant quality loss	Fits: 125K context · ~1560 tok/s
0.4 GB	0.8 GB	—	0.1 GB	2-bit · i-quant extra small — Importance-weighted 2-bit, extra small	Fits: 125K context · ~1524 tok/s
0.4 GB	0.8 GB	—	0.1 GB	2-bit · i-quant small — Importance-weighted 2-bit, small	Fits: 125K context · ~1524 tok/s
0.4 GB	0.8 GB	—	0.1 GB	2-bit · i-quant medium — Importance-weighted 2-bit, medium	Fits: 125K context · ~1489 tok/s
0.5 GB	0.8 GB	—	0.1 GB	3-bit · i-quant — Importance-weighted 3-bit	Fits: 125K context · ~1456 tok/s
0.5 GB	0.8 GB	—	0.1 GB	3-bit · i-quant extra small — Importance-weighted 3-bit, extra small	Fits: 125K context · ~1456 tok/s
0.5 GB	0.8 GB	—	0.1 GB	2-bit · small K-quant — 2-bit small K-quant	Fits: 125K context · ~1456 tok/s
0.5 GB	0.8 GB	—	0.1 GB	3-bit · i-quant small — Importance-weighted 3-bit, small	Fits: 125K context · ~1424 tok/s
0.5 GB	0.8 GB	—	0.1 GB	3-bit · i-quant medium — Importance-weighted 3-bit, medium	Fits: 125K context · ~1424 tok/s
0.5 GB	0.8 GB	—	0.1 GB	3-bit small — 3-bit small quantization	Fits: 125K context · ~1424 tok/s
0.5 GB	0.8 GB	—	0.1 GB	2-bit · K-quant — 2-bit quantization with K-quant improvements	Fits: 125K context · ~1424 tok/s
0.5 GB	0.8 GB	—	0.1 GB	3-bit medium — 3-bit medium quantization	Fits: 125K context · ~1394 tok/s
0.5 GB	0.8 GB	—	0.1 GB	4-bit · legacy — 4-bit legacy quantization	Fits: 125K context · ~1394 tok/s
0.5 GB	0.8 GB	—	0.1 GB	4-bit · i-quant compact — Importance-weighted 4-bit, compact	Fits: 125K context · ~1365 tok/s
0.5 GB	0.8 GB	—	0.1 GB	3-bit large — 3-bit large quantization	Fits: 125K context · ~1365 tok/s
0.5 GB	0.8 GB	—	0.1 GB	4-bit small — 4-bit small quantization	Fits: 125K context · ~1337 tok/s
0.5 GB	0.8 GB	—	0.2 GB	4-bit · legacy+ — 4-bit legacy quantization with offset	Fits: 125K context · ~1337 tok/s
0.5 GB	0.8 GB	—	0.1 GB	4-bit · i-quant non-linear — Importance-weighted 4-bit, non-linear	Fits: 125K context · ~1337 tok/s
0.5 GB	0.8 GB	—	0.2 GB	5-bit · legacy — 5-bit legacy quantization	Fits: 125K context · ~1310 tok/s
0.5 GB	0.8 GB	—	0.2 GB	4-bit medium — 4-bit medium quantization — most popular sweet spot	Fits: 125K context · ~1310 tok/s
0.5 GB	0.9 GB	—	0.2 GB	5-bit medium — 5-bit medium quantization — good quality/size tradeoff	Fits: 125K context · ~1260 tok/s
0.5 GB	0.9 GB	—	0.2 GB	5-bit small — 5-bit small quantization	Fits: 125K context · ~1260 tok/s
0.6 GB	0.9 GB	—	0.2 GB	6-bit, very good — 6-bit quantization, very good quality	Fits: 125K context · ~1191 tok/s
0.6 GB	0.9 GB	—	0.2 GB	8-bit, near-lossless — 8-bit quantization, near-lossless	Fits: 125K context · ~1111 tok/s
0.8 GB	1.2 GB	—	0.5 GB	16-bit · bfloat — Brain floating point 16 — preferred for training	Fits: 125K context · ~799 tok/s

Selected Q4_K_M

4-bit medium quantization

0.6GBat 16K context

Weights: 0.1 GB
KV cache: 0.2 GB
Overhead: 0.3 GB

Fits RTX 4090 (reference) · 23.4 GB free · ~1310 tok/s

≥50 tok/s needs ≥38 GB/s · RTX 5090 ~2330, RTX 3090 Ti ~1310

Which Hardware Can Run LFM2.5 230M?

Q4_K_M · 0.5 GB

LFM2.5 230M (Q4_K_M) requires 0.5 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 1+ GB is recommended. Using the full 128K context window can add up to 1.5 GB, bringing total usage to 2.0 GB. 52 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti. 163 devices with unified memory can run LFM2.5 230M, 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 · 0.6 GB · ~354 tok/s · 125K0.6~354125K8 GB tier →

12 GB · RTX 3060 12GBQ4_K_M · 0.6 GB · ~468 tok/s · 125K0.6~468125K12 GB tier →

16 GB · RTX 5060 Ti 16GBQ4_K_M · 0.6 GB · ~582 tok/s · 125K0.6~582125K16 GB tier →

24 GB · RTX 4090referenceQ4_K_M · 0.6 GB · ~1310 tok/s · 125K0.6~1310125K24 GB tier →

Mac 32 GB · Mac Mini M4Q4_K_M · 0.6 GB · ~168 tok/s · 125K0.6~168125KMacs →

Mac 64 GB · Mac Studio M4 MaxQ4_K_M · 0.6 GB · ~764 tok/s · 125K0.6~764125KMacs →

Alternatives to LFM2.5 230M

LFM2.5 family · 14 models →

Similar models

2tok/s on RTX 4090

Model	Score	VRAM	tok/s	Why it is similar
Tiny MixtralTitanML	—	0.5 GB	~2329	same size
T5 Paraphrase PawsVamsi	—	0.2 GB	~4368	same size

Fine-tunes & variants(1)community builds of LFM2.5 230M

LFM2.5 230M ONNX

Specifications

Publisher: Liquid AI
Family: LFM2.5
Parameters: 230M
Architecture: Lfm2ForCausalLM
Context Length: 128,000 tokens
Vocabulary Size: 65,536
Release Date: 2026-06-24
License: Other

Resources

Frequently Asked Questions

How much VRAM does LFM2.5 230M need?

LFM2.5 230M requires 0.5 GB of VRAM at Q4_K_M, or 0.8 GB at BF16. Full 128K context adds up to 1.5 GB (2.0 GB total).

VRAM = Weights + KV Cache + Overhead

Weights = 230M × 4.8 bits ÷ 8 = 0.1 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 ≈ 1.9 GB (at full 128K context)

VRAM usage by quantization

Q4_K_M

0.5 GB

Q4_K_M + full context

2.0 GB

Learn more about VRAM estimation →

What's the best quantization for LFM2.5 230M?

For LFM2.5 230M, Q4_K_M (0.5 GB) offers the best balance of quality and VRAM usage. Q5_0 (0.5 GB) provides better quality if you have the VRAM. The smallest option is IQ2_XXS at 0.4 GB.

VRAM requirement by quantization

IQ2_XXS

0.4 GB~53%

Q2_K_S

0.5 GB~71%

Q4_0

0.5 GB~85%

Q5_0

0.5 GB~90%

Q4_K_M ★

0.5 GB~89%

BF16

0.8 GB~100%

★ Recommended — best balance of quality and VRAM usage.

Learn more about quantization →

Can I run LFM2.5 230M on a Mac?

Yes — MacBook Air 13" M3 (8 GB) and 145 other Macs can run LFM2.5 230M. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 0.5 GB of usable unified memory (RAM minus macOS overhead).

Can I run LFM2.5 230M locally?

Yes — LFM2.5 230M can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.

How fast is LFM2.5 230M?

At Q4_K_M, LFM2.5 230M can reach ~9600 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~1310 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 ÷ 0.5 × 0.65 = ~10400 tok/s

Estimated speed at Q4_K_M (0.5 GB)

NVIDIA B200

~10400 tok/s

NVIDIA GeForce RTX 4090

~1310 tok/s

NVIDIA B300

~10400 tok/s

AMD Instinct MI350X

~9600 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 LFM2.5 230M?

At Q4_K_M, the download is about 0.14 GB. The full-precision BF16 version is 0.46 GB. The smallest option (IQ2_XXS) is 0.06 GB.

Which GPUs can run LFM2.5 230M?

52 consumer GPUs can run LFM2.5 230M 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 LFM2.5 230M?

163 devices with unified memory can run LFM2.5 230M at Q4_K_M (0.5 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.