LFM3 1.7B Base
LFM3 1.7B Base is an open-weights language model from Liquid AI, released October 26, 2026. It has 1.7B parameters (dense). The context window is 256K tokens.
Check it against your own PC →
Can you run LFM3 1.7B Base locally?
The smallest common GPU it runs on is 6 GB of VRAM (with 32 GB of RAM): Q6_K, fully on GPU, roughly 75 tokens per second. Estimates assume 32 GB of DDR5 RAM, a 16K context and llama.cpp-style GGUF quantization. Enter your exact hardware in the live app.
| GPU memory | Verdict | Best quant | VRAM | RAM | tok/s |
|---|---|---|---|---|---|
| 6 GB VRAM | Runs on GPU | Q6_K (≈ lossless) | 2.2 GB | – | ~75 |
| 8 GB VRAM | Runs on GPU | Q6_K (≈ lossless) | 2.2 GB | – | ~100 |
| 12 GB VRAM | Runs on GPU | Q6_K (≈ lossless) | 2.2 GB | – | ~130 |
| 16 GB VRAM | Runs on GPU | Q6_K (≈ lossless) | 2.2 GB | – | ~170 |
| 24 GB VRAM | Runs on GPU | Q6_K (≈ lossless) | 2.2 GB | – | ~280 |
| 32 GB VRAM | Runs on GPU | Q6_K (≈ lossless) | 2.2 GB | – | ~430 |
Ready-made quantizations on Hugging Face: Mungert/LFM3-1.7B-GGUF · saidutta69/LFM3-1.7B-Base-heretic · mradermacher/LFM3-1.7B-Base-GGUF · mradermacher/LFM3-1.7B-Base-i1-GGUF · VelociQ8/LFM3-1.7B-Base-Q8A64-GGUF · efeisot/lfm3-1.7b-1rco-2mix-gguf
Benchmark results
Epoch AI has not published results for LFM3 1.7B Base yet. New models are usually evaluated within days to weeks.