# Liquid AI releases LFM3, six open-weight 2B-class models for on-device text, vision and audio

Canonical: https://theagenticweekly.com/news/liquid-ai-lfm3-on-device-model-family · Evidence: Reported · First seen: 2026-11-01 00:18 UTC · Source published: not stated by the source · Brief published: 2026-11-01 00:23 UTC · Updated: 2026-11-01 00:23 UTC · Author: Agentic Weekly

> Liquid AI released LFM3 on October 27, 2026: six open-weight models of about 2B parameters branched from one LFM3-1.7B-Base backbone pre-trained on 22T tokens. Liquid AI says it found "LFM3-1.7B to outperform our previous LFM2.5-2.6B while using 35% fewer parameters".

Models in this brief: [LFM3-1.7B](https://theagenticweekly.com/models/lfm3-1-7b)

## What changed

- The family covers text (LFM3-1.7B), vision (LFM3-VL-2B), audio (LFM3-Audio-2B) and omni (LFM3-Omni-2B), plus two decision models specialized from the vision and omni models.
- All six branch from LFM3-1.7B-Base, pre-trained on 22T tokens on a new hybrid architecture.
- The text model has a 256K-token context window and fits in about 1 GB at 4-bit precision.
- Weights are on Hugging Face under the LFM Open License v1.0, with GGUF, MLX, ONNX, FP8 and NVFP4 builds.

## In the source's words

> "Built for the edge: the text model fits in about 1 GB at 4-bit precision."
> (Liquid AI: LFM3 announcement, https://www.liquid.ai/blog/lfm3)

> "LFM3-1.7B to outperform our previous LFM2.5-2.6B while using 35% fewer parameters"
> (Liquid AI: LFM3 announcement, https://www.liquid.ai/blog/lfm3)

## What we have not verified

- Performance comparisons are Liquid AI's own; we did not benchmark LFM3.
- We did not run the models on a laptop or phone.

## Timeline

- Source published: not stated by the source
- First seen by our watcher: 2026-11-01 00:18 UTC (via Hugging Face blog (editorial release sweep))
- Brief published: 2026-11-01 00:23 UTC

## Sources

- [Liquid AI blog: LFM3](https://www.liquid.ai/blog/lfm3) (primary source) (fetched 2026-10-31)
- [Hugging Face blog: LFM3](https://huggingface.co/blog/LiquidAI/lfm3) (primary source) (fetched 2026-10-31)
- [Hugging Face model card: LFM3-1.7B](https://huggingface.co/LiquidAI/LFM3-1.7B) (primary source) (fetched 2026-10-31)

## Related

- [LFM3-1.7B: model facts and sources](https://theagenticweekly.com/models/lfm3-1-7b)
- [All models: compare price, context and evidence](https://theagenticweekly.com/models)
- [All news briefs](https://theagenticweekly.com/news)
- [This week's releases](https://theagenticweekly.com/this-week)
- [New AI models this week](https://theagenticweekly.com/models/this-week)
- [AI model releases by month](https://theagenticweekly.com/models/releases)
- [Best LLM for coding](https://theagenticweekly.com/best/llm-for-coding)

## How to cite

Agentic Weekly, "Liquid AI releases LFM3, six open-weight 2B-class models for on-device text, vision and audio", published 2026-11-01 00:23 UTC, https://theagenticweekly.com/news/liquid-ai-lfm3-on-device-model-family