lfm3

50 Downloads Updated an hour ago

LFM3-1.7B is a tri-hybrid model for local agents and specialized applications.

tools thinking 1.7b

ollama run lfm3

curl http://localhost:11434/api/chat \
  -d '{
    "model": "lfm3",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

from ollama import chat

response = chat(
    model='lfm3',
    messages=[{'role': 'user', 'content': 'Hello!'}],
)
print(response.message.content)

import ollama from 'ollama'

const response = await ollama.chat({
  model: 'lfm3',
  messages: [{role: 'user', content: 'Hello!'}],
})
console.log(response.message.content)

Applications

Claude Code

Claude Code ollama launch claude --model lfm3

OpenCode

OpenCode ollama launch opencode --model lfm3

Hermes Agent

Hermes Agent ollama launch hermes --model lfm3

OpenClaw

OpenClaw ollama launch openclaw --model lfm3

Models

Name

5 models

Size / Usage

Context

Input

lfm3:latest

1.7GB · 256K context window · Text · an hour ago

lfm3:latest

1.7GB

256K

Text

lfm3:1.7b

1.7GB · 256K context window · Text · an hour ago

lfm3:1.7b latest

1.7GB

256K

Text

Readme

LFM3-1.7B is a text model for local agents and specialized applications, based on a new tri-hybrid architecture.

Agentic training: post-trained for agentic harnesses such as Hermes Agent, Pi, and OpenClaw, with a 256K-token context window.
Designed for the edge: at 4-bit precision, it takes about 1 GB of memory, so it fits on any phone or laptop.
Made to be specialized: adapt the model to your domains, formats, and tools via quick single-GPU fine-tuning.

Model details

The LFM3 family includes base, text, vision, audio, omni, and decision models.

Property	Specification
Checkpoint	Post-trained, text-only language model
Parameters	1.73B
Architecture	24 layers: 13 gated short convolutions, 5 sliding-window attention, 6 global attention
Context length	256K tokens (input + output)
Vocabulary	128,000 tokens
Languages	English, Arabic, Chinese, French, German, Italian, Japanese, Korean, Portuguese, Spanish, Vietnamese, Thai, Indonesian, Hindi, Russian, Polish
License	LFM Open License v1.0

LFM3-1.7B is designed for local agents, tool calling, structured outputs, and domain-specific fine-tuning. Use retrieval or search for knowledge-heavy tasks. It is not recommended for agentic coding or complex multi-step reasoning due to its small size.

LFM3-1.7B is a **text model for local agents and specialized applications**, based on a new tri-hybrid architecture.

* **Agentic training:** post-trained for agentic harnesses such as Hermes Agent, Pi, and OpenClaw, with a 256K-token context window.
* **Designed for the edge:** at 4-bit precision, it takes about 1 GB of memory, so it fits on any phone or laptop.
* **Made to be specialized:** adapt the model to your domains, formats, and tools via quick single-GPU fine-tuning.

# Model details
The LFM3 family includes base, text, vision, audio, omni, and decision models.

| Property | Specification |
|:--- |:---:|
| Checkpoint | Post-trained, text-only language model |
| Parameters | 1.73B |
| Architecture |        24 layers: 13 gated short convolutions, 5 sliding-window attention, 6 global attention
| Context length|       256K tokens (input + output)
| Vocabulary |  128,000 tokens
| Languages |   English, Arabic, Chinese, French, German, Italian, Japanese, Korean, Portuguese, Spanish, Vietnamese, Thai, Indonesian, Hindi, Russian, Polish
| License |    LFM Open License v1.0
LFM3-1.7B is designed for local agents, tool calling, structured outputs, and domain-specific fine-tuning. Use retrieval or search for knowledge-heavy tasks. It is not recommended for agentic coding or complex multi-step reasoning due to its small size.

Paste, drop or click to upload images (.png, .jpeg, .jpg, .svg, .gif)