LFM2.5 2.6B GGUF
LFM2.5 2.6B GGUF is a text-generation GGUF checkpoint from the LiquidAI/LFM2.5-2.6B-GGUF repository. It is part of the LFM2.5 hybrid model family, described in the model card as designed for on-device deployment and built on the LFM2 architecture with extended pre-training and reinforcement learning. The card lists multilingual support for Arabic, Chinese, English, French, German, Hindi, Indonesian, Italian, Japanese, Korean, Polish, Portuguese, Russian, Spanish, Thai, and Vietnamese. The repository provides GGUF files for llama.cpp, including a Quantization-Aware Distillation QAD Q4_0 GGUF and a post-training-quantized Q4_0 GGUF, plus FP32 QAD source weights under qad/ for Transformers fine-tuning and experimentation. The license is lfm1.0. No performance, hardware, compatibility, or numerical requirements are stated here beyond the supplied facts.
What do you want to do?
Choose a task. Get the software, files, settings, and steps to run it.
Open a guide for its software, hardware requirements, and setup steps.
At a glance
- Architecture
- unknown
- License
- lfm1.0
- Disk space
- 1.6 GB
- Software
- llama.cpp, Transformers
Will it run on your machine?
Save your machine to see a personalized rating and its reasoning.
Add my hardwareBest for
- Local text generation using GGUF and llama.cpp
- Conversational use via llama-cli with the supplied example parameters
- Fine-tuning and experimentation with the FP32 QAD source weights in Transformers
- Multilingual text generation for the listed languages
- On-device deployment scenarios described by the model card
Tradeoffs
- The model card states that published QAD results apply to the Q4_0 GGUF, and direct FP32/BF16 inference or other quantization formats may behave differently.
- The license is lfm1.0 with license_name other; review the LICENSE file for terms.
- The model card does not provide performance benchmarks, hardware requirements, compatibility guarantees, or numerical metrics for this GGUF repository.
- The QAD Q4_0 GGUF is distinct from the post-training-quantized Q4_0 GGUF, although both use the GGUF Q4_0 format.
- The supplied model facts do not establish testing, validation, or suitability for any particular deployment beyond what the model card states.
Ways to run it
llama.cpp · Windows, macOS, Linux
Repository-specific command in publisher documentation; confirm dependencies and hardware in the source.
llama-cli -hf LiquidAI/LFM2.5-2.6B-GGUF --conversation --temp 0.1 --top-k 50 --repeat-penalty 1.1Transformers · See official guide
Repository-specific command in publisher documentation; confirm dependencies and hardware in the source.
from transformers import AutoModelForCausalLM, AutoTokenizer
repo_id = "LiquidAI/LFM2.5-2.6B-GGUF"
tokenizer = AutoTokenizer.from_pretrained(repo_id, subfolder="qad")
model = AutoModelForCausalLM.from_pretrained(
repo_id, subfolder="qad", dtype="auto", device_map="auto"
)
inputs = tokenizer.apply_chat_template(
[{"role": "user", "content": "What is 2 + 2?"}],
tokenize=True, add_generation_prompt=True, return_dict=True, return_tensors="pt",
).to(model.device)
outputs = model.generate(**inputs, max_new_tokens=128)
print(tokenizer.decode(outputs[0, inputs["input_ids"].shape[-1]:], skip_special_tokens=True))Explore its uses
Keep exploring
Bonsai 2 · 27B
A compressed 27B-class reasoning model with publisher-provided GGUF packs and a dedicated llama.cpp fork for CUDA, Metal, and CPU.
DeepSeek V4.1 Flash
A multimodal reasoning model with a compressed key-value cache, published as open weights.
Qwen3.8
Explore 4 versions of Qwen3.8. Compare model sizes, packages, hardware requirements, and available setups.