LibreAI / AI Agents / Vision-Agents
// agents · GetStream/Vision-Agents

Vision-Agents

Stream's open Vision Agents — build real-time voice and vision agents with any model or video provider.

★ 8.2k stars⑂ 692 forksApache-2.0updated 2026-10
▦ visual — this project's README ships no embeddable screenshot,
so here's a placeholder. The guide below is complete regardless.
// project facts

Project facts

Vision-Agents is an open-source ai agents project by GetStream (GetStream/Vision-Agents on GitHub). It has 8,154 GitHub stars and 692 forks, is released under the Apache-2.0 license, and was last updated on GitHub on 8 October 2026.

GitHub stars
8,154
Forks
692
License
Apache-2.0
Last updated on GitHub
8 October 2026
Category
AI Agents
Listing
Full install guide
Official site
Official site
// how it works

How it works

  1. Clarify your goal

    Pick your video provider, the models you will combine (for example a detector plus an LLM), and get Stream credentials if you use Stream's edge network.

  2. Get the project

    Clone GetStream/Vision-Agents from GitHub (or use the install path in the run steps below) and skim its README for the parts relevant to your goal.

  3. Install and configure

    Add vision-agents with uv, optionally with the extras bundle for Stream, OpenAI, ElevenLabs and Deepgram integrations.

  4. Run your first task

    Follow the quickstart to join a call with a basic agent — the repo's golf-coach example shows the pattern — and have it respond to what it sees and hears.

// features

Why people love it

Real-time video AI

Built for live video: combine YOLO, Roboflow and others with Gemini or OpenAI in real time.

Low-latency edge

Uses Stream's edge network with fast joins and very low audio/video latency, while remaining open to other video providers.

Native model APIs

Work through native SDK methods from OpenAI, Gemini and Claude rather than a lowest-common-denominator wrapper.

Many client SDKs

SDKs for React, Android, iOS, Flutter, React Native and Unity powered by the same realtime stack.

Voice and vision together

Speech-to-text, text-to-speech and vision integrations (extras include ElevenLabs and Deepgram) for agents that listen and speak as well as watch.

Worked examples

Example agents, including a golf coach, show full patterns for perception-plus-LLM video agents.

Advertisement
// run it

Run it yourself

  1. Clone the repository

    Get the source from the official GitHub repository for Vision-Agents.

    git clone https://github.com/GetStream/Vision-Agents.git
    cd Vision-Agents
  2. Install with uv

    Install the core package as the README's getting-started step 1 shows.

    uv add vision-agents
  3. Add integrations and credentials

    Install the extras bundle for the providers you will use, then get a Stream API key and follow the quickstart to build your first agent.

    uv add "vision-agents[getstream, openai, elevenlabs, deepgram]"  # then add your Stream API credentials — see README
// use cases

What you can build

Advertisement
// faq

Questions, answered

Who maintains Vision-Agents?

GetStream (Stream), through the GetStream/Vision-Agents repository, with the site at visionagents.ai.

Is Vision-Agents free to use?

Yes — the project is open source under the Apache-2.0 license as recorded for this listing. See the license terms for any conditions on redistribution.

Does Vision-Agents need a paid API key?

You need credentials for the services you use — for Stream, the README notes a free API key with monthly participant minutes; model providers such as OpenAI or Gemini use your own keys.

How does Vision Agents compare to calling a vision model API on uploaded clips?

Upload-and-analyse APIs are offline and high-latency. Vision Agents is built for live sessions — streaming video, sub-second joins and realtime responses — which suits interactive agents but means running a realtime stack, not a batch job.

How do I get started with Vision-Agents?

Clone the repository (or use the documented installer), follow the install and configure steps on this page, and run the smallest real task first — follow the quickstart to join a call with a basic agent — the repo's golf-coach example shows the pattern — and have it respond to what it sees and hears.

// keep exploring

Similar tools

AI Agents

LocalAI

LocalAI is the open-source AI engine. Run any model - LLMs, vision, voice, image, video - on any hardware. No GPU required.

★ 49.4k starsMIT
View guide →
Agents

qwen-audio-agent

Real-time voice runtime that keeps AI agents talking, working, and present.

★ 2.8k starsApache-2.0
View guide →
AI Agents

Tutti

Tutti — a real-time shared workspace where people and multiple AI agents build in tune, sharing context, files, apps and tasks

★ 3.8k starsApache-2.0
View guide →
AI Agents

graphiti

Real-time knowledge graphs that give agents a memory with structure.

★ 31.5k starsApache-2.0
View guide →
AI Agents

harness-sdk

An open-source SDK for production agents on any model, any cloud.

★ 8.7k starsApache-2.0
View guide →
AI Agents

vocode-core

Build voice-based LLM agents with a modular, open-source toolkit.

★ 3.8k starsMIT
View guide →
// alternatives

Alternatives in AI Agents

The most-starred other AI Agents projects in the LibreAI catalogue — the realistic shortlist if Vision-Agents is not the right fit.

Get Vision-Agents

Free and open — grab the source, star the repo, and start creating.