# LFM3-1.7B

Model tracker · checked 2026-10-29 · https://dbggr.dev/models/lfm3-1-7b/

Open-weight 1.7B text model for local agents, tool calling and structured output

LFM3-1.7B is a fast open-weight model from Liquid AI, released 27 October 2026. At Liquid AI it costs $0.08 per million input tokens and $0.20 per million output tokens, with a 262K-token context window.

- Made by: Liquid AI
- Released: 2026-10-27
- Status: Generally available, new
- Tier: Fast
- Weights: Open
- Memory to run: About 5 GB (BF16)
- Context window: 262,144 tokens
- Input: text
- Docs: https://docs.liquid.ai/lfm/models/lfm3-1.7b

## LFM3-1.7B price at Liquid AI

US dollars per million tokens. Blended is 3 input tokens for every output token, one number to compare on.

| Provider | Input | Output | Cached input | Blended | Batch | Checked |
|---|---|---|---|---|---|---|
| Liquid AI | $0.08 | $0.20 | $0.02 | $0.11 | – | [2026-10-29](https://www.liquid.ai/pricing) |

- Liquid AI: Liquid API. Context is 256K tokens for input and output combined. No batch price listed.

## LFM3-1.7B benchmarks

| Benchmark | Score | What it measures | Reported by |
|---|---|---|---|
| GPQA Diamond | 63.4% | graduate-level science questions answered correctly | [Liquid AI](https://huggingface.co/LiquidAI/LFM3-1.7B) |

Scores reported by a model's maker aren't independently checked.

## Can you run LFM3-1.7B yourself?

Yes. Liquid AI publishes the weights on Hugging Face as [LiquidAI/LFM3-1.7B](https://huggingface.co/LiquidAI/LFM3-1.7B), under LFM Open License v1.0: 1.73B parameters, 4 GB in BF16.

Running it takes about 5 GB of GPU memory: a single GPU with at least 5 GB of memory, or a Mac with comfortably more than 5 GB. That's the weight files plus 20% for working memory with short inputs; long contexts need more, and fitting in memory doesn't make it fast.

Liquid AI also publishes a smaller MLX 4-bit version, [LiquidAI/LFM3-1.7B-MLX-4bit](https://huggingface.co/LiquidAI/LFM3-1.7B-MLX-4bit), at 1 GB: about 2 GB to run, so a single GPU with at least 2 GB of memory, or a Mac with comfortably more than 2 GB.

## Models like LFM3-1.7B

| Model | Input | Output | Context | Weights |
|---|---|---|---|---|
| LFM3-1.7B | $0.08 | $0.20 | 262K | Open |
| [Qwen3.5 9B](https://dbggr.dev/models/qwen3-5-9b/) | $0.10 | $0.15 | 262K | Open |
| [Ministral 3 3B](https://dbggr.dev/models/ministral-3-3b/) | $0.10 | $0.10 | 256K | Open |
| [NVIDIA Nemotron 3.5 Lightning 30B A3B](https://dbggr.dev/models/nvidia-nemotron-3-5-lightning-30b-a3b/) | $0.06 | $0.16 | 262K | Open |
| [Ministral 3 8B](https://dbggr.dev/models/ministral-3-8b/) | $0.15 | $0.15 | 256K | Open |
| [Llama 3.3 70B](https://dbggr.dev/models/llama-3-3-70b/) | $0.10 | $0.32 | 131K | Open |

## FAQ

### How much does LFM3-1.7B cost?

LFM3-1.7B costs $0.08 per million input tokens and $0.20 per million output tokens at Liquid AI, and $0.02 per million cached input tokens. Prices are list prices in US dollars, last checked 29 October 2026.

### What is LFM3-1.7B's context window?

LFM3-1.7B takes up to 262,144 tokens in one request.

### Is LFM3-1.7B open weight?

Yes. Liquid AI publishes the weights, so you can run LFM3-1.7B yourself or at any host that serves it.

### Can I run LFM3-1.7B locally?

LFM3-1.7B's weights are 4 GB in BF16, so it needs about 5 GB of GPU memory: a single GPU with at least 5 GB of memory, or a Mac with comfortably more than 5 GB. Liquid AI also publishes a MLX 4-bit version that needs about 2 GB.

### Where can I use LFM3-1.7B?

LFM3-1.7B is available from Liquid AI.