LFM2.5-350M-GGUF
LFM2 is a new generation of hybrid models developed by Liquid AI, specifically designed for edge AI and on-device deployment. It sets a new standard in terms of quality, speed, and memory efficiency.
Params
350 M
Context
128,000
Downloads 30d
165 K
Likes
110
Download history
tracking started — chart appears after 7 days of snapshots (5 recorded)165 K downloads in the last 30 days
Can you run it?
Estimated VRAM at 8K context unless noted. Pick your hardware to see the verdict per quantization.
| File | Quant | Size | Est. VRAM | Verdict on RTX 4090 · 24 GB |
|---|---|---|---|---|
| LFM2.5-350M-QAD-Q4_0.gguf | Q4_0 | 0.2 GB | 0.8 GB | ✅ Runs comfortably |
| LFM2.5-350M-Q4_K_M.gguf | Q4_K_M | 0.2 GB | 0.8 GB | ✅ Runs comfortably |
| LFM2.5-350M-Q5_K_M.gguf | Q5_K_M | 0.3 GB | 0.8 GB | ✅ Runs comfortably |
| LFM2.5-350M-Q6_K.gguf | Q6_K | 0.3 GB | 0.9 GB | ✅ Runs comfortably |
| LFM2.5-350M-Q8_0.gguf | Q8_0 | 0.4 GB | 1.0 GB | ✅ Runs comfortably |
| LFM2.5-350M-BF16.gguf | GGUF | 0.7 GB | 1.3 GB | ✅ Runs comfortably |
| model.safetensors | fp16 | 1.4 GB | 2.1 GB | ✅ Runs comfortably |
| model.safetensors (bf16, full) | bf16 + 125K ctx | 1.4 GB | 2.9 GB | ✅ Runs comfortably |
Estimate: file size × 1.1 + KV cache at 8K + 0.5 GB overhead. Not a benchmark — how we calculate this.
Run it
copy-paste, exact tags checked against the Hub$ ollama run lfm2-5-350m-gguf # pin the quantization explicitly $ ollama run lfm2-5-350m-gguf-q4_0
$ huggingface-cli download LiquidAI/LFM2.5-350M-GGUF-GGUF \
LFM2.5-350M-QAD-Q4_0.gguf --local-dir .
$ llama-cli -m LFM2.5-350M-QAD-Q4_0.gguf \
-c 8192 -ngl 99 -t 8 --color
$ curl -s https://aimodelscomparison.com/api/v1/models/lfm2-5-350m-gguf
{
"hf_id": "LiquidAI/LFM2.5-350M-GGUF",
"params_b": 0.35,
"context_length": 128000,
"license": { "id": "other", "commercial": "unknown" },
"downloads_30d": 164626,
"vram_estimates": [
{ "quant": "GGUF", "gb": 1.3 },
{ "quant": "Q4_0", "gb": 0.8 }
],
"updated_at": "2026-11-06T01:00:25Z"
}
Specifications
- Parameters
- 350 M
- Context length
- 128,000
- Licence
- other
- First seen on the Hub
- 2026-03-25
- Base model
- LFM2.5-350M
- Training datasets
- undisclosed
- Added to our catalog
- 2026-11-06
Family
Base model and the most-downloaded derivatives in the catalog.
Compare with any text-generation model