Open Poe AI
https://github.com/Anil-matcha/Open-Poe-AI- Category
- AI Tools
- Rank
- No. 1541Tools index
Previous survey · No. 1565 ·
- Pricing
- Open Source
- Type
- APP
- Builder
- Anil-matcha
- GitHub
- 21 stars
- Date
About
Self-hosted Poe alternative — chat with GPT, Claude, Gemini, Llama and other LLMs from one UI using your own API keys.
What it does
Open Poe AI organizes reusable agents around a persona, capability, profile, and conversation history. Its browser interface supports text conversations alongside image, video, and audio generation. Users can browse personal, template, and featured agents, build agents visually, and share profiles within an instance.
Why it's ranked here
The scope is more useful than a basic chat console: one agent library covers conversations and several media formats. A visual builder, stored histories, templates, suggestions, and profiles make agents reusable. However, the project calls itself early work in progress, and its backend forwards requests to another configured provider rather than supplying generation directly.
What's good
Agent definitions package prompts, capabilities, profiles, and histories into reusable units. The library separates personal, template, and featured agents, with search across names and descriptions. The same interaction model accepts conversational replies or generated images, videos, and audio. The backend also exposes creation, editing, likes, suggestions, skills, and health checks through REST endpoints.
Tradeoffs
Deployment requires separate Node and Python services plus credentials for an external provider. Several development defaults assume local ports, including frontend rewrites, server-side requests, and a CORS allowlist. The supplied client wrapper uses a mock user identity. Provider independence is architectural rather than automatic, since every useful chat or generation request still depends on a compatible configured upstream service.
How to use it well
Use it when a team wants a self-managed catalogue of repeatable personas and media workflows, with visual editing and per-conversation history. It fits internal experimentation, reusable support agents, and mixed text or media generation. Budget time for configuring the upstream service, credentials, networking, and identity integration. It does not replace the model or media provider that performs generation.
Technical notes+
The private npm workspace in package.json spans client, server, and packages/agents. client/next.config.mjs transpiles the shared ai-agent package, permits images from cdn.muapi.ai, and rewrites /api traffic to localhost:8000. server/app/main.py creates the FastAPI application, restricts CORS to localhost:3000, mounts the proxy router under /api, and exposes root and health routes. packages/agents/src/index.js exports the main agent UI, create and edit pages, profile, theme provider, themes, and a server fetch helper. client/context/fetchAgentData.js makes uncached requests directly to the local backend, while client/components/AgentClientWrapper.js supplies a mock user context.
Observed
- License
- MIT
- Languages
- JavaScript for the Next.js client and shared React package; Python for the FastAPI backend
- Packaging
- Private npm workspace with client, server, and packages directories; backend dependencies install from a pip requirements file
- Interfaces
- Browser interface, reusable React component package, and REST API
- Platform support
- Self-hosted on infrastructure capable of running Node and Python
- Architecture
- Next.js frontend, shared React agent components, and a FastAPI proxy backend
Read from README.md, package.json, packages/agents/postcss.config.js, packages/agents/tailwind.config.js, packages/agents/src/index.js, packages/agents/src/utils/server.js, client/next.config.mjs, client/postcss.config.mjs, server/app/main.py, client/app/page.js, client/app/layout.js, client/context/fetchAgentData.js, client/components/AgentClientWrapper.js, client/app/agents/page.js, client/app/agents/loading.js.
What it can do
Chat with GPT models
Text prompts and API keys → AI-generated responses from GPT
Chat with Claude models
Text prompts and API keys → AI-generated responses from Claude
Chat with Gemini models
Text prompts and API keys → AI-generated responses from Gemini
Chat with Llama models
Text prompts and API keys → AI-generated responses from Llama
Switch between different LLMs
User selection and configured API keys → Active connection to chosen AI model
Manage multiple AI conversations
User prompts across different chat sessions → Organized conversation threads with different models
Self-host AI chat interface
Server setup and API key configuration → Private AI chat application instance
Tags
Tech Stack
Media
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.