> ## Documentation Index
> Fetch the complete documentation index at: https://liquidai-liren-deprecate-leap-sdk.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# INTERACTIVE QUICKSTART WIDGET

# Interactive Quickstart Widget

## Complete list of Deployment platforms:

* Laptop with Transformers
* Laptop with Ollama
* Laptop with llama.cpp
* Macbook with MLX
* iOS with llama.cpp
* Android with llama.cpp
* Cloud with vLLM
* Browser with Transformers.js

## Complete list of use cases:

* Chat completions
* Vision understanding
* Audio & Transcription
* Code generation
* Text embeddings
* Function Calling & Agents

## Complete list of models:

* LiquidAI/LFM2-8B-A1B
* LiquidAI/LFM2-2.6B
* LiquidAI/LFM2-1.2B
* LiquidAI/LFM2-700M
* LiquidAI/LFM2-350M
* LiquidAI/LFM2-VL-3B
* LiquidAI/LFM2-VL-1.6B
* LiquidAI/LFM2-VL-450M
* LiquidAI/LFM2-Audio-1.5B
* LiquidAI/LFM2-ColBERT-350M

## Compatibility matrix between use cases, deployment platforms and models

* LFM2 models
  * Models:
    * LiquidAI/LFM2-8B-A1B
      * Description: On-device MoE with 8B parameters, 1B active, comparable to 3-4B dense models and faster than Qwen3-1.7B. Code and knowledge capabilities are significantly improved compared to LFM2-2.6B.

    * LiquidAI/LFM2-2.6B
      * Description: Mid-size model for balanced performance

    * LiquidAI/LFM2-1.2B
      * Description: Compact model for general use

    * LiquidAI/LFM2-700M
      * Description: Smaller efficient model

    * LiquidAI/LFM2-350M
      * Description: Ultra-lightweight for edge devices

  * Use cases:
    * Text completions
    * Code generation
    * Function calling and Agents

  * Deployment platforms:
    * Transformers
    * Ollama
    * llama.cpp
    * MLX
    * iOS
    * Android
    * vLLM

* LFM2-VL
  * Models:
    * LiquidAI/LFM2-VL-3B
      * Description: Lightweight 3B vision-language model with enhanced visual reasoning and fine-grained perception, built on the LFM2 backbone for efficient multimodal understanding at variable resolutions.
    * LiquidAI/LFM2-VL-1.6B
      * Description: Compact 1.6B vision-language model balancing strong multimodal capabilities with efficient inference, built on the LFM2 backbone for practical visual understanding at variable resolutions.
    * LiquidAI/LFM2-VL-450M
      * Description: Ultra-lightweight 450M vision-language model optimized for resource-constrained deployments, delivering essential multimodal understanding with minimal compute requirements and efficient on-device inference

  * Use cases:
    * Vision understanding

  * Deployment platforms:
    * Transformers
    * Ollama
    * llama.cpp
    * iOS
    * Android

* LFM2-Audio
  * Models
    * LiquidAI/LFM2-Audio-1.5B
      * Description:
  * Use cases:
    * Audio & Transcription
  * Deployment platforms:
    * liquid-audio library
    * llama.cpp

* LiquidAI/LFM2-ColBERT-350M
  * Description: late interaction retriever with excellent multilingual performance. It allows you to store documents in one language (for example, a product description in English) and retrieve them in many languages with high accuracy.
  * Use cases:
    * Text Embeddings
  * Deployment platforms:
    * Transformers

* Models to exclude
  * LiquidAI/LFM2-1.2B-Extract
  * LiquidAI/LFM2-350M-Extract
  * LiquidAI/LFM2-350M-ENJP-MT
  * LiquidAI/LFM2-1.2B-RAG
  * LiquidAI/LFM2-1.2B-Tool
  * LiquidAI/LFM2-350M-Math
  * LiquidAI/LFM2-350M-PII-Extract-JP

## Tutorial templates

* Use case: Chat completions
* Deployment platform: Laptop with Transformers
* Model: {MODEL}
* Models:
  * LiquidAI/LFM2-8B-A1B
  * LiquidAI/LFM2-2.6B
  * LiquidAI/LFM2-1.2B
  * LiquidAI/LFM2-700M
  * LiquidAI/LFM2-350M
  * LiquidAI/LFM2-VL-3B
  * LiquidAI/LFM2-VL-1.6B
  * LiquidAI/LFM2-VL-450M
  * LiquidAI/LFM2-Audio-1.5B
  * LiquidAI/LFM2-ColBERT-350M

## Steps

## Step 1. Install Python dependencies

```
pip install transformers torch
```

## Step 2. Run inference with the `pipeline()` interface

```
from transformers import pipeline

# Load model
generator = pipeline("text-generation", "LiquidAI/LFM2-1.2B", device_map="auto")

# Generate
messages = [{"role": "user", "content": "What is machine learning?"}]
response = generator(messages, max_new_tokens=256)
print(response[0]["generated_text"][-1]["content"])
```