Poe

Overview

Poe (Platform for Open Exploration) is Quora's AI chat app that puts many AI models behind a single interface. It launched in December 2022 and opened to the public in February 2023, founded by Quora CEO Adam D'Angelo. Rather than subscribing to each provider, users chat with models from OpenAI, Anthropic, Google, Meta, DeepSeek, and others, plus image, video, and audio generators, all from one account on web, iOS, Android, macOS, and Windows.

What it does

The core experience is multi-model chat: pick a model, ask, and get an answer, or compare models side by side. On top of that, Poe lets anyone build custom bots, either no-code prompt bots or code-backed server bots, and publish them to Poe's audience, with creator monetization via price per message or subscription revenue share. Multi-bot and group chats let people mix models and creator bots in one conversation, and group chats (launched November 2025) support up to 200 people across more than 200 text, image, video, and audio models. Image, video, and audio generation models sit alongside the text models so users create media in the same app. For developers, an OpenAI-compatible API exposes all Poe models and bots, with tool calling enabled across OpenAI, Anthropic, and Google models.

Autonomy note

Treat Poe as an assistant-grade platform: it responds when asked rather than acting on its own. The API's tool-calling support means developers can build agentic workflows on Poe, but that autonomy lives in the developer's application, not in Poe's default chat experience.

Integrations & setup

Available on the web at poe.com plus iOS, Android, macOS, and Windows apps, with chats synced across devices. The OpenAI-compatible API (managed at poe.com/api_key) plugs Poe into existing tools such as Cline, Roo Code, Continue, Cursor, the llm CLI, n8n, and Codex CLI, so a single Poe subscription can power coding assistants and automation that expect an OpenAI-style endpoint. Poe is model-agnostic by design, aggregating frontier and open-source models from many labs.

Pricing

Freemium. A free tier gives limited daily access to models. Paid subscriptions start at $4.99/month and scale up (reported up to higher enterprise-style tiers) for higher usage caps, compute-point allowances, and access to premium models. Poe also surfaces transparent USD per-token costs on model pages and lets users and developers buy additional credits to go beyond subscription limits. Exact prices, point allowances, and limits change; check the official subscription page.

Best for / not for

Best for users who want one place to use and compare many AI models (and to generate images, video, and audio) without juggling separate subscriptions, for creators who want to build and distribute bots, and for developers who want OpenAI-compatible access to many models through one key. Less suited to buyers who need a hands-off autonomous agent out of the box; Poe is an aggregation and chat layer, and any agentic behavior has to be built on its API.

Alternatives

ChatGPT, Claude, and Google Gemini are the single-provider chat assistants Poe aggregates and competes with; Perplexity competes on cited web search; OpenRouter is the closest developer-facing multi-model API aggregator. Poe's distinguishing angle is the combination of consumer multi-model chat, a creator-bot marketplace, and a unified API.