LFM2.5-2.6B-GGUF
LFM2.5 is a new family of hybrid models designed for on-device deployment. It builds on the LFM2 architecture with extended pre-training and reinforcement learning.
Params
2.6 B
Context
131,072
Downloads 30d
187 K
Likes
397
Download history
daily snapshots · 89 days
▲ 797 K in the last 30 days (81.0%)
972 K187 K
Oct 12Oct 22Nov 1Nov 10
1.2 M187 K
Aug 14Sep 12Oct 12Nov 10
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 |
|---|---|---|---|---|
| LFM2.5-2.6B-QAD-Q4_0.gguf | Q4_0 | 1.6 GB | 2.6 GB | ✅ Runs comfortably |
| LFM2.5-2.6B-Q4_K_M.gguf | Q4_K_M | 1.7 GB | 2.7 GB | ✅ Runs comfortably |
| LFM2.5-2.6B-Q5_K_M.gguf | Q5_K_M | 1.9 GB | 3.0 GB | ✅ Runs comfortably |
| LFM2.5-2.6B-Q6_K.gguf | Q6_K | 2.2 GB | 3.3 GB | ✅ Runs comfortably |
| LFM2.5-2.6B-Q8_0.gguf | Q8_0 | 2.9 GB | 4.1 GB | ✅ Runs comfortably |
| LFM2.5-2.6B-BF16.gguf | GGUF | 5.4 GB | 6.8 GB | ✅ Runs comfortably |
| model.safetensors | fp16 | 10.8 GB | 12.8 GB | ✅ Runs comfortably |
| model.safetensors (bf16, full) | bf16 + 128K ctx | 10.8 GB | 18.6 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 lfm2-5-2-6b-gguf # pin the quantization explicitly $ ollama run lfm2-5-2-6b-gguf-q4_0
$ huggingface-cli download LiquidAI/LFM2.5-2.6B-GGUF-GGUF \
LFM2.5-2.6B-QAD-Q4_0.gguf --local-dir .
$ llama-cli -m LFM2.5-2.6B-QAD-Q4_0.gguf \
-c 8192 -ngl 99 -t 8 --color
$ curl -s https://aimodelscomparison.com/api/v1/models/lfm2-5-2-6b-gguf
{
"hf_id": "LiquidAI/LFM2.5-2.6B-GGUF",
"params_b": 2.60,
"context_length": 131072,
"license": { "id": "other", "commercial": "unknown" },
"downloads_30d": 186819,
"vram_estimates": [
{ "quant": "GGUF", "gb": 6.8 },
{ "quant": "Q4_0", "gb": 2.6 }
],
"updated_at": "2026-09-23T01:00:05Z"
}
Specifications
- Parameters
- 2.6 B
- Context length
- 131,072
- Licence
- other
- First seen on the Hub
- 2026-08-01
- Base model
- LFM2.5-2.6B
- Training datasets
- undisclosed
- Added to our catalog
- 2026-08-14
Family
Base model and the most-downloaded derivatives in the catalog.
Compare with any text-generation model