# LFM3-1.7B

LFM3-1.7B is Liquid AI's latest language model in our dataset, released Oct 27, 2026. Open weights (LFM Open License v1.0). Liquid AI's 1.7B open-weights text model for on-device agents, the text member of the six-model LFM3 family announced Oct 27, 2026, with a 256K-token context window. Verified against vendor sources on Oct 31, 2026.

Primary source (source of record; Agentic Weekly is the fast independent index): https://www.liquid.ai/blog/lfm3

Last verified: 2026-10-31 · Evidence: Reported · Canonical: https://theagenticweekly.com/models/lfm3-1-7b · JSON: https://theagenticweekly.com/api/models/lfm3-1-7b

## Key facts

| Fact | Value | Source | Label | Anchor |
|---|---|---|---|---|
| Vendor | Liquid AI |  | editorial | https://theagenticweekly.com/models/lfm3-1-7b#fact-vendor |
| Released | Oct 27, 2026 | https://www.liquid.ai/blog/lfm3 | vendor-reported, confirmed | https://theagenticweekly.com/models/lfm3-1-7b#fact-release-date |
| Context window | 256,000 tokens (256K). Stated as 256K tokens. | https://www.liquid.ai/blog/lfm3 | vendor-reported, confirmed | https://theagenticweekly.com/models/lfm3-1-7b#fact-context-tokens |
| Input | text | https://huggingface.co/LiquidAI/LFM3-1.7B | vendor-reported, confirmed | https://theagenticweekly.com/models/lfm3-1-7b#fact-input-modalities |
| Output | text | https://huggingface.co/LiquidAI/LFM3-1.7B | vendor-reported, confirmed | https://theagenticweekly.com/models/lfm3-1-7b#fact-output-modalities |
| Parameters | 1.7B total, Hybrid LFM3 architecture; pre-trained on 22T tokens | https://www.liquid.ai/blog/lfm3 | vendor-reported, confirmed | https://theagenticweekly.com/models/lfm3-1-7b#fact-parameters |
| Licence | LFM Open License v1.0 (open weights) | https://huggingface.co/LiquidAI/LFM3-1.7B | vendor-reported, confirmed | https://theagenticweekly.com/models/lfm3-1-7b#fact-license |

## Vendor claims and neutral signals

- Liquid AI says the text model fits in about 1 GB at 4-bit precision and outperforms its LFM2.5-2.6B while using 35% fewer parameters. Vendor claim; we did not run it.

## Notes

- GGUF, MLX, ONNX, FP8 and NVFP4 builds are on Hugging Face.

## What we tested

Nothing yet. We have not run LFM3-1.7B ourselves, so every fact here is Reported from the sources below.

## Not verified, left out

- The vision, audio, omni and two decision models in the LFM3 family are not tracked as separate entries yet.
- Benchmark scores were not captured.

## Alternatives to LFM3-1.7B

Models from other vendors that share at least one of our groups, from each model's sourced facts (not tested by us).

- [Nemotron 3.5 Bolt E4B (4GB)](https://theagenticweekly.com/models/nemotron-3-5-bolt-e4b) (NVIDIA): Both in: Open weights, Runs on your own hardware, Small and fast · 64K context · released Oct 23, 2026
- [Gemma 4](https://theagenticweekly.com/models/gemma-4) (Google): Both in: Open weights, Runs on your own hardware, Small and fast · 256K context · released Mar 31, 2026
- [Mistral Small 4](https://theagenticweekly.com/models/mistral-small-4) (Mistral AI): Both in: Open weights, Runs on your own hardware, Small and fast · $0.15 in / $0.6 out per 1M tokens · 256K context · released Mar 16, 2026
- [Ministral 3 14B](https://theagenticweekly.com/models/ministral-3-14b) (Mistral AI): Both in: Open weights, Runs on your own hardware, Small and fast · $0.2 in / $0.2 out per 1M tokens · 256K context · released Dec 2, 2025
- [SenseNova-Loop-9B-Preview](https://theagenticweekly.com/models/sensenova-loop-9b-preview) (SenseNova): Both in: Open weights, Runs on your own hardware · 262K context · released Oct 30, 2026
- [MiniMax M3.1 Flash](https://theagenticweekly.com/models/minimax-m3-1-flash) (MiniMax): Both in: Open weights, Small and fast · $0.15 in / $0.6 out per 1M tokens · 1M context · released Oct 19, 2026
- [Best open-weight LLMs: how to choose](https://theagenticweekly.com/best/open-weight-llms)

## LFM3-1.7B FAQ

**When was LFM3-1.7B released?**

Liquid AI released LFM3-1.7B on Oct 27, 2026. Source: https://www.liquid.ai/blog/lfm3

**How large is the LFM3-1.7B context window?**

256,000 tokens. Source: https://www.liquid.ai/blog/lfm3

**Is LFM3-1.7B open weights?**

Yes. The weights are published under LFM Open License v1.0. Source: https://huggingface.co/LiquidAI/LFM3-1.7B

**Has Agentic Weekly tested LFM3-1.7B?**

Not yet. Every fact on this page is Reported: read from the linked vendor pages, not measured by us.

## Sources

- [Liquid AI blog: LFM3](https://www.liquid.ai/blog/lfm3)
- [Hugging Face model card](https://huggingface.co/LiquidAI/LFM3-1.7B)
- [Agentic Weekly brief](https://theagenticweekly.com/news/liquid-ai-lfm3-on-device-model-family)

Data licence: CC BY 4.0. Data: Agentic Weekly (https://theagenticweekly.com), CC BY 4.0.