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

# Liquid Nanos

> A library of low-latency, task-specific models fine-tuned on Liquid's multimodal LFM base models. Nanos deliver high accuracy on narrow tasks while remaining small enough to deploy on-device or serve economically at high volume.

<div className="use-cases" style={{marginTop: '-0.5rem'}}>
  <CardGroup cols={2}>
    <Card title="Extraction" icon="brackets-curly">
      Structured data extraction from unstructured documents into JSON.
    </Card>

    <Card title="Summarization" icon="align-left">
      Meeting transcripts, document summaries, and content distillation.
    </Card>

    <Card title="RAG & Retrieval" icon="magnifying-glass">
      Context-grounded Q\&A and fast multi-language document retrieval.
    </Card>

    <Card title="Translation" icon="language">
      Low-latency bidirectional translation for short-to-medium text.
    </Card>
  </CardGroup>
</div>

<Note>Many Nanos require specific prompting formats to work correctly. See each model's page for usage guidelines.</Note>

<h2>LFM2.5 Nanos   <Badge shape="rounded" icon="circle-check" color="green" size="lg">Latest release</Badge></h2>

<CardGroup cols={2}>
  <Card title="LFM2.5-Embedding-350M" href="/lfm/models/lfm25-embedding-350m">
    350M · <Badge shape="pill" color="orange">Retrieval</Badge>

    Dense bi-encoder for the smallest, fastest multilingual vector index.
  </Card>

  <Card title="LFM2.5-ColBERT-350M" href="/lfm/models/lfm25-colbert-350m">
    350M · <Badge shape="pill" color="orange">Retrieval</Badge>

    Late-interaction retriever for higher accuracy and reranking quality.
  </Card>

  <Card title="LFM2.5-VL-1.6B-Extract" href="/lfm/models/lfm25-vl-1.6b-extract">
    1.6B · <Badge shape="pill" color="blue">Vision Extraction</Badge>

    Extract user-defined fields from images into structured JSON.
  </Card>

  <Card title="LFM2.5-VL-450M-Extract" href="/lfm/models/lfm25-vl-450m-extract">
    450M · <Badge shape="pill" color="blue">Vision Extraction</Badge>

    Compact image-to-JSON extraction model for edge workflows.
  </Card>
</CardGroup>

## LFM2 Nanos

<CardGroup cols={3}>
  <Card title="LFM2-350M-PII-Extract-JP" href="/lfm/models/lfm2-350m-pii-extract-jp">
    350M · <Badge shape="pill" color="blue">Extraction</Badge>

    Japanese PII detection into structured JSON.
  </Card>

  <Card title="LFM2-2.6B-Transcript" href="/lfm/models/lfm2-2.6b-transcript">
    2.6B · <Badge shape="pill" color="purple">Summarization</Badge>

    Private, on-device meeting summarization from transcripts.
  </Card>

  <Card title="LFM2-350M-ENJP-MT" href="/lfm/models/lfm2-350m-enjp-mt">
    350M · <Badge shape="pill" color="red">Translation</Badge>

    Near real-time bidirectional Japanese/English translation.
  </Card>

  <Card title="LFM2-350M-Math" href="/lfm/models/lfm2-350m-math">
    350M · <Badge shape="pill" color="yellow">Reasoning</Badge>

    Tiny reasoning model for math problem solving.
  </Card>

  <Card title="LFM2-1.2B-Extract" href="/lfm/models/lfm2-1.2b-extract">
    1.2B · <Badge shape="pill" color="gray">Deprecated</Badge>

    Use LFM2.5-1.2B-Instruct for structured extraction.
  </Card>

  <Card title="LFM2-350M-Extract" href="/lfm/models/lfm2-350m-extract">
    350M · <Badge shape="pill" color="gray">Deprecated</Badge>

    Use LFM2.5-350M for compact extraction workflows.
  </Card>

  <Card title="LFM2-1.2B-RAG" href="/lfm/models/lfm2-1.2b-rag">
    1.2B · <Badge shape="pill" color="gray">Deprecated</Badge>

    Use LFM2.5-1.2B-Instruct with LFM2.5 retrieval models.
  </Card>

  <Card title="LFM2-ColBERT-350M" href="/lfm/models/lfm2-colbert-350m">
    350M · <Badge shape="pill" color="gray">Deprecated</Badge>

    Use LFM2.5-ColBERT-350M instead.
  </Card>

  <Card title="LFM2-1.2B-Tool" href="/lfm/models/lfm2-1.2b-tool">
    1.2B · <Badge shape="pill" color="gray">Deprecated</Badge>

    Use LFM2.5-1.2B-Instruct for tool calling instead.
  </Card>
</CardGroup>

<Panel>
  * [Liquid Playground](https://playground.liquid.ai/chat?model=cmk0wefde000204jp2knb2qr8)
  * [HuggingFace Collections](https://huggingface.co/LiquidAI/collections)
  * [OpenRouter API](https://openrouter.ai/liquid)
  * [LEAP Model Library](https://leap.liquid.ai/models)
</Panel>