LFM3 by Liquid AI
LFM3 is Liquid AI's new family of six open-weight models around 2B parameters (text, vision, audio, omni and two decision models) built on one shared backbone, with a text model that has a 256K context window and fits in about 1 GB at 4-bit.
Why it matters
For developers who want capable AI agents running on phones, laptops, edge devices or robots instead of the cloud. The models are small enough to run locally and to fine-tune on a single GPU for a specific job.
Highlights
- Six 2B-class models (text, vision, audio, omni and two decision models) all branched from LFM3-1.7B-Base, pre-trained on 22T tokens.
- LFM3-1.7B is post-trained for agent harnesses such as Hermes Agent, Pi and OpenClaw, with a 256K-token context window.
- Liquid AI says LFM3-1.7B outperforms its earlier LFM2.5-2.6B with 35% fewer parameters, and is about 25% faster than LFM2-1.2B with 8K tokens in context.unverified
- Day-one support for llama.cpp, MLX and vLLM, plus launch partners including AMD, Arm, Intel (OpenVINO), NVIDIA, OpenRouter, Qualcomm and Vercel.
- Every model can be fine-tuned on a single GPU; cookbook notebooks are on GitHub.
Taken from the project's own README, docs or launch page. Claims about speed or accuracy are the maker's unless stated otherwise.
Worth knowing: Companies at or above $10M in annual revenue aren't licensed for commercial use under the LFM Open License. Benchmark results are Liquid AI's own.
Quick facts
| Pricing | Weights free to download; Liquid AI API billed separately (rates not checked)not verified |
|---|---|
| Free or paid | Open weights (free) |
| License | LFM Open License v1.0 (Apache-style; commercial use only for organizations under $10M annual revenue)not verified |
| Language | Model |
| Platform | Hugging Face weights; llama.cpp, MLX, vLLM; phones, laptops, Raspberry Pi 5, GPUs and NPUs per Liquid AI |
| GitHub stars | n/a (not a GitHub project) |
| Maker | Liquid AI |
Links
- Official site: https://www.liquid.ai/blog/lfm3
- Docs: https://docs.liquid.ai/lfm/getting-started/welcome
Related ranking
See how established tools compare in Best AI Agents That Do Tasks for You. New releases aren't scored until we review them against the rubric.
What we couldn't verify
- highlights[2] (performance and speed claims are Liquid AI's)
- quick_facts.pricing (API rates not checked)
- quick_facts.license (checked on LFM3-1.7B; other LFM3 models assumed to match)
Checked 2026-10-28 from the model's own pages. Prices and star counts change, so confirm on the official site. Listing here is not a ranking or an endorsement.
One email a week on AI that saves business owners time
Every Friday: one tool pick with the business use case, what changed in tools you already use, and the week's AI news in plain English. No spam.
Free. Unsubscribe any time. See our privacy policy.