Mixpeek breaks every video, image, and audio file into structured features
your agents can search, reason over, and trust.

Docs · Get Started · Quickstart · Blog ·

Mixpeek is multimodal infrastructure for AI agents. Upload video, images, audio, and documents — Mixpeek automatically extracts features (faces, objects, transcripts, embeddings, structured metadata) and indexes them into searchable collections. Your agent queries a single endpoint and gets structured results back.

Index → Upload files to buckets. Mixpeek runs feature extraction automatically — faces, objects, transcripts, embeddings, and structured metadata all get indexed.

Search → Build retrieval pipelines. Semantic search, face search, object search, transcript search — chain them into multi-stage retrievers exposed as a single endpoint.

Integrate → Wire Mixpeek into your agent as a LangChain tool, an MCP server, or a direct REST call.

pip install mixpeek

from mixpeek import Mixpeek

mx = Mixpeek(api_key="YOUR_API_KEY")

# Upload a video
mx.buckets.upload(bucket_id="my-bucket", file_path="video.mp4")

# Search across all extracted features
results = mx.retrievers.execute(
    retriever_id="my-retriever",
    inputs={"query_text": "person wearing a red jacket"},
    limit=10,
)

Also available as:

• JavaScript SDK: npm install mixpeek
• MCP Server: Connect Claude, Cursor, or any MCP-compatible agent
• REST API: POST https://api.mixpeek.com/v1/retrievers/{id}/execute
• CLI: mixpeek --version (included in the Python SDK)

Each extracted feature becomes an independently searchable document. A single video can produce hundreds of documents — one per face, one per transcript segment, one per scene.

┌─────────────────────────────────────────────────────────┐
│                      Your Agent                         │
│         (LangChain · MCP · REST · SDK · CLI)            │
└──────────────────────┬──────────────────────────────────┘
                       │
┌──────────────────────▼──────────────────────────────────┐
│                   API Layer                              │
│            FastAPI + Celery Workers                      │
│   Buckets · Collections · Retrievers · Webhooks         │
└──────────────────────┬──────────────────────────────────┘
                       │
┌──────────────────────▼──────────────────────────────────┐
│                 Engine Layer                             │
│                Ray Serve Cluster                         │
│   SigLIP · ArcFace · Whisper · Gemini · E5 · LayoutLM  │
└──────────────────────┬──────────────────────────────────┘
                       │
┌──────────────────────▼──────────────────────────────────┐
│                    Storage                               │
│       MongoDB · Qdrant · Redis · S3-compatible          │
└─────────────────────────────────────────────────────────┘

Object Storage: AWS S3, Google Cloud Storage, Azure Blob, Cloudflare R2, Backblaze B2, Supabase, Wasabi, Tigris, Mux, Box

Agent Frameworks: LangChain, MCP (Model Context Protocol), OpenAI Function Calling, direct REST

Data Warehouses: Snowflake, Databricks

• Video understanding — Search surveillance footage by face, scene, or spoken word
• Content moderation — Detect brand logos, faces, and unsafe content across media libraries
• Document intelligence — Extract structured data from scanned PDFs, invoices, and forms
• Media asset management — Find the exact frame across millions of hours of video
• E-commerce — Visual similarity search, product matching, catalog enrichment

• Documentation
• API Reference
• Studio — Visual dashboard for managing namespaces, collections, and retrievers
• Status