Skip to content

Text model · LFM

LF LFM2 350M: RAM and VRAM requirements

LFM2 350M needs about 1.4 GB to run at Q4_K_M (the Q4_K_M GGUF file is ~0.45 GB to download; KV cache and overhead add the rest), or about 1.3 GB at Q8_0. The lightest hardware that runs it is Apple M1 (8GB).

LFM family · 0.354B params · released Nov 2025.

Shopping for hardware? See what runs LFM2 350M →

License LFM Open License v1.0 · Conditional ↓ 24.6K/mo ♥ 248 on HuggingFace
Q4_K_M GGUF
0.45 GB
Q8_0 GGUF
0.35 GB
Memory @ Q4 (4k)
~1.4 GB
Context
128 k

Q4_K_M and Q8_0 are GGUF download (file) sizes. Memory @ Q4 includes the KV cache and overhead, so it runs larger than the file on disk.

Will it run on your device?

LFM2 350M runs on 40 of 40 tracked devices at Q4_K_M.

40 run well 0 tight fit 0 too small
Fit check Q4_K_M
Yes, it runs fast
needs 1.4 GB usable 10.5 GB

Memory breakdown

Weights (Q4_K_M)0.45 GB
+
KV cache (4k)0.1 GB
+
Overhead0.8 GB
=
Total1.4 GB

How context length changes it

4k context ~1.4 GB 32k context ~2.4 GB 128k context ~5.9 GB

LFM2 350M is small enough that the context window is the thing to size for: 128k needs ~5.9 GB versus ~1.4 GB at 4k, so pick the window you actually use.

Speed drops too: every token re-reads the KV cache, so at 128k context LFM2 350M generates at roughly ~9% of its short-context speed. How this is estimated.

Quantization sizes

GGUF quantson disk
Quantization Size on disk
Q2_K 0.1 GB est
Q3_K_M 0.2 GB est
Q4_K_M (default) 0.45 GB
Q5_K_M 0.3 GB est
Q6_K 0.3 GB est
Q8_0 0.35 GB
FP16 0.711 GB

Lower quant = smaller and faster, slightly lower quality. Q4_K_M is the common default.

Run it

llama.cpp
$ llama-cli -hf LiquidAI/LFM2-350M-GGUF:Q4_K_M
LM Studio
$ lms get LiquidAI/LFM2-350M-GGUF

Which devices can run LFM2 350M?

Same job, different size

Pick by what fits your memory: step down to free up VRAM, or step up if you have headroom.

Similar models

Head-to-head

FAQ

How much VRAM or RAM does LFM2 350M need?

At Q4_K_M, LFM2 350M needs about 1.4 GB (weights ~0.45 GB + KV cache + overhead) at a 4k context. At Q8_0 budget ~1.3 GB.

What is the Q4_K_M GGUF file size of LFM2 350M?

The Q4_K_M GGUF file is about 0.45 GB to download, and the Q8_0 GGUF is about 0.35 GB. That is the weights file on disk; to run it you also need room for the KV cache and overhead, so budget ~1.4 GB of memory at Q4_K_M.

Can LFM2 350M run on a laptop?

Yes, LFM2 350M fits on a 16 GB machine at Q4_K_M and runs on Apple Silicon or a 12 GB+ GPU comfortably.

Can I use LFM2 350M commercially?

Conditionally. LFM Open License: free under $10M revenue; larger companies need a commercial license.

Understand the numbers

Short guides to the ideas behind LFM2 350M's memory and quant figures.

Exact params: 354,483,968. Hybrid architecture: 10 double-gated short-range LIV convolution blocks + 6 GQA blocks. Context 128K from config.json max_position_embeddings. LFM Open License: free for academic/research and companies with revenue under $10M; larger enterprises must license commercially. No official ollama.com/library tag; community sam860/LFM2:350m exists. Paper arXiv:2511.23404 released November 2025. Q4_K_M 229 MB is the standard (non-hip-optimized) file.

Sources

Last validated 2026-08-03. Memory figures are estimates. See methodology.