LFM2.5 230M

LFM2.5 is a family of hybrid models designed for on-device deployment. It builds on the LFM2 architecture with extended pre-training and reinforcement learning.

• Our most compact model yet: 230M parameters that punch above their weight, bringing real capability to the tightest memory and compute budgets.
• Fast edge inference: Best throughput from low-cost CPUs to production GPUs, running at 213 tok/s decode speed on Galaxy S25 Ultra and 42 tok/s on a Raspberry Pi 5.
• Built for agentic tasks: Distilled from LFM2.5-350M and refined with multi-stage reinforcement learning, making it well-suited for tool use and data extraction.

Find more information about LFM2.5-230M in our blog post.

🗒️ Model Details

LFM2.5-230M is a general-purpose text-only model with the following features:

• Number of parameters: 230M
• Number of layers: 14 (8 double-gated LIV convolution blocks + 6 GQA blocks)
• Training budget: 19T tokens
• Context length: 32,768 tokens
• Vocabulary size: 65,536
• Knowledge cutoff: Mid-2024
• Languages: English, Arabic, Chinese, French, German, Italian, Japanese, Korean, Portuguese, Spanish
• Generation parameters:
  • temperature: 0.1
  • top_k: 50
  • repetition_penalty: 1.05

We recommend using it for data extraction and lightweight on-device agentic pipelines. It is not recommended for reasoning-heavy workloads such as advanced math, code generation, or creative writing.

Chat Template

LFM2.5 uses a ChatML-like format. See the Chat Template documentation for details. Example:

You can use tokenizer.apply_chat_template() to format your messages automatically.

Tool Use

LFM2.5 supports function calling in four steps:

1. Function definition: Provide the list of tools as a JSON object in the system prompt, or use tokenizer.apply_chat_template() with tools=....
2. Function call: By default, LFM2.5 writes Pythonic function calls (a Python list between <|tool_call_start|> and <|tool_call_end|> special tokens), as the assistant answer. You can override this behavior by asking the model to output JSON function calls in the system prompt.
3. Function execution: Execute the call and return the result with the tool role.
4. Final answer: LFM2.5 interprets the tool output and returns a plain-text answer addressing the original prompt.

See the Tool Use documentation for the full guide. Example:

🏃 Inference

LFM2.5 is supported by many inference frameworks. See the Inference documentation for the full list.

Quick start with Transformers (compatible with transformers>=5.0.0):

from transformers import AutoModelForCausalLM, AutoTokenizer, TextStreamer

model_id = "LiquidAI/LFM2.5-230M"
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    device_map="auto",
    dtype="bfloat16",

)
tokenizer = AutoTokenizer.from_pretrained(model_id)
streamer = TextStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True)

prompt = "What is C. elegans?"

input_ids = tokenizer.apply_chat_template(
    [{"role": "user", "content": prompt}],
    add_generation_prompt=True,
    return_tensors="pt",
    tokenize=True,
)["input_ids"].to(model.device)

output = model.generate(
    input_ids,
    do_sample=True,
    temperature=0.1,
    top_k=50,
    repetition_penalty=1.05,
    max_new_tokens=512,
    streamer=streamer,
)

🔧 Fine-Tuning

We recommend fine-tuning LFM2.5 for your specific use case to achieve the best results.

📊 Performance

Benchmarks

CPU Inference

GPU Inference

📬 Contact

• Got questions or want to connect? Join our Discord community
• If you are interested in custom solutions with edge deployment, please contact our sales team.

Citation

@article{liquidAI2026230M,
  author = {Liquid AI},
  title = {LFM2.5-230M: Built to Run Anywhere},
  journal = {Liquid AI Blog},
  year = {2026},
  note = {www.liquid.ai/blog/lfm2-5-230m},
}

@article{liquidai2025lfm2,
  title={LFM2 Technical Report},
  author={Liquid AI},
  journal={arXiv preprint arXiv:2511.23404},
  year={2025}
}