Model reference · open weights
LFM3 is an open-weight language model from LiquidAI. LFM3-1.7B-NVFP4 (NVFP4) weighs 1.8 GB; the smallest configuration that runs it is RTX 3060 12 GB.
For self-hosting, compare this build's requirements below with the GPU you plan to rent.
Summary of the LiquidAI/LFM3-1.7B-GGUF model card, 2026-10-29. The estimate below is for another build of the family.
What it is
| Released by | LiquidAI |
|---|---|
| Released | 2026-10-24 |
| VRAM | 1.8 GB for the weights |
What it runs on
| Card | Requests at once | Context max | Memory | |
|---|---|---|---|---|
| 8K each | 32K each | |||
| RTX 3060 12 GB FP4 without its speed-up here | 25 | 6 | 228K | 11.6 GB |
| RTX 4060 Ti 16 GB FP4 without its speed-up here | 35 | 9 | all 256K | 15.4 GB |
| RTX 3090 24 GB FP4 without its speed-up here | 57 | 15 | all 256K | 23.4 GB |
| RTX 4090 24 GB FP4 without its speed-up here | 57 | 15 | all 256K | 23.4 GB |
| RTX 5090 32 GB | 78 | 21 | all 256K | 31.0 GB |
| L40S 48 GB FP4 without its speed-up here | 114 | 31 | all 256K | 44.0 GB |
| A100 80 GB FP4 without its speed-up here | 208 | 57 | all 256K | 78.2 GB |
| H100 80 GB FP4 without its speed-up here | 178 | 49 | all 256K | 78.1 GB |
| RTX PRO 6000 Blackwell 96 GB | 217 | 60 | all 256K | 93.8 GB |
| DGX Spark (GB10) 128 GB unified | 250 | 69 | all 256K | 107 GB |
| H200 141 GB FP4 without its speed-up here | 326 | 91 | all 256K | 138 GB |
| B200 180 GB | 420 | 105 | all 256K | 176 GB |
| Requests at once | 8K tokens each | 32K tokens each |
|---|---|---|
| 1 | 2.9 GB | 3.8 GB |
| 5 | 4.3 GB | 9.1 GB |
| 8 | 5.4 GB | 13.0 GB |
| 16 | 8.3 GB | 23.6 GB |
| 32 | 14.1 GB | 44.7 GB |
| 64 | 25.7 GB | 86.9 GB |
One card, with vLLM's small-card settings.
Builds
| Build | Params | Precision | Weights | Smallest setup |
|---|---|---|---|---|
| LFM3-1.7B-GGUF ↗ 7 builds: Q4_0 1.0 GB … F16 3.5 GB |
GGUF | 1.1 GB | RTX 3060 12 GB | |
| LFM3-1.7B ↗ | 1.7B | BF16 | 3.5 GB | RTX 3060 12 GB |
| LFM3-1.7B-NVFP4 (above) ↗ | 1.2B | NVFP4 | 1.8 GB | RTX 3060 12 GB |
| LFM3-1.7B-MLX-4bit ↗ | 1.7B | MLX | 973 MB | RTX 3060 12 GB |
From the model card
src="https://huggingface.co/LiquidAI/LFM3-1.7B/resolve/main/assets/liquid-ai-banner.webp" alt="Liquid AI" style="width: 100%; max-width: 100%; height: auto; display: block; margin: 0;" />
GGUF version of LFM3-1.7B for local inference on CPUs and GPUs.
Find architecture and usage details in the original model card.
Recommended sampling: temperature 0.1, top_k 50, repeat penalty 1.1. For math and science: temperature 1.0, top_p 0.95, top_k 20.
ollama run lfm3), for macOS. Requires Ollama version v0.40.12 or newer.Add a quantization tag to pick a file.
llama-cli -hf LiquidAI/LFM3-1.7B-GGUF:Q4_K_M \
--temp 0.1 --top-k 50 --repeat-penalty 1.1
For math and science, use --temp 1.0 --top-p 0.95 --top-k 20 instead. More options in the llama.cpp guide.
The Q4_0 build in this repository is the Quantization-Aware Distillation (QAD) checkpoint,
LFM3-1.7B-QAD-Q4_0.gguf.
It was trained to recover the accuracy lost to Q4_0 quantization.
Example usage with llama.cpp:
llama-cli -hf LiquidAI/LFM3-1.7B-GGUF \
--hf-file LFM3-1.7B-QAD-Q4_0.gguf \
--temp 0.1 --top-k 50 --repeat-penalty 1.1
Quoted from the model card on Hugging Face. The full card is behind the Hugging Face link above.
How it works