Last updated: 2026-10-28
Launched in February 2023 by Quora, Poe is an AI model aggregator that gives one chat interface to 200+ official models from OpenAI, Anthropic, Google, and Meta, plus more than 1 million user-created bots and an OpenAI-compatible API for developers.
About Poe
Quora founded Poe internally in 2022 and launched it publicly in February 2023, then raised a $75M round in January 2024 led by Andreessen Horowitz. The platform has a two-fold mission: give everyday users one interface to compare AI products from multiple providers, and let developers build and monetise custom bots without deep technical expertise.
Poe's creator program lets bot builders publish either no-code prompt bots or full API-backed server bots for public use and earn a share of the revenue those bots generate, a path most single-vendor chat apps do not offer at all. The app syncs conversations across web, iOS, Android, macOS and Windows.
Billing runs on compute points rather than flat per-model fees, so one subscription covers every provider in the catalogue. Worth knowing before comparing plans: Poe's subscription plans have no marketing names. Each plan card is labelled only by its point allowance, so there is no Starter, Pro, Premium or Enterprise tier to ask for, and the largest plan is the top consumer allowance rather than a business contract.
Plan prices are readable only on Poe's subscription page, poe.com/subscription_plans; the full list, monthly and yearly, sits in the pricing section of this page.
The data posture deserves a careful read, because Poe is a broker rather than a first-party model host: whatever you type travels on to the outside labs and bot builders behind each bot. The governance section on this page quotes the policy itself, including the group-chat exception to the identity masking.
Who should skip Poe: if you live in one vendor's models, a first-party plan such as ChatGPT, Claude or Gemini goes deeper than a broker can, for example OpenAI's own Codex agent. Research-heavy users who want cited answers are better served by Perplexity, and developers who only need API access to many models can skip the chat app and the points arithmetic with OpenRouter.
Pricing
Poe is free to start with a daily compute-point allowance. It sells compute points, and its plans carry no names; each plan card is labelled by its allowance alone. Billed monthly, "10 thousand points/day" is $4.99, "660 thousand points/month" $19.99, "1.65 million points/month" $49.99, "3.3 million points/month" $99.99 and "8.25 million points/month" $249.99.
Billed yearly, Poe's default view with a SAVE 17% badge, the same five plans cost $49.99, $199.99, $499.99, $999.99 and $2,499.99 per year, shown as $4.17, $16.67, $41.67, $83.33 and $208.33 per month. Every paid plan opens each bot's full context length, up to 2M tokens according to Poe, and add-on points cost $30 per 1M points. There is no Enterprise plan: the largest tier is the top consumer subscription.
Poe notes that "your final price may vary based on regional taxes and currency conversion", and subscriptions are managed and cancelled through Stripe's customer portal. poe.com/pricing and poe.com/plans both returned 404 on 28 October 2026, so poe.com/subscription_plans is the only public price surface. To pay per model instead of in points, compare OpenRouter; for one vendor's flagship on a flat plan, ChatGPT or Claude.
| Tier | Monthly price | What it includes |
|---|---|---|
| Free | Free | Daily compute-point allowance; premium models not included |
| 10 thousand points/day | $4.99/mo | 10,000 compute points per day. About $4.99/month billed monthly, or $49.99/year billed yearly ($4.17/month equivalent, SAVE 17%) |
| 660 thousand points/month | $19.99/mo | 660,000 compute points per month. About $19.99/month billed monthly, or $199.99/year billed yearly ($16.67/month equivalent, SAVE 17%) |
| 1.65 million points/month | $49.99/mo | 1,650,000 compute points per month. About $49.99/month billed monthly, or $499.99/year billed yearly ($41.67/month equivalent, SAVE 17%) |
| 3.3 million points/month | $99.99/mo | 3,300,000 compute points per month. About $99.99/month billed monthly, or $999.99/year billed yearly ($83.33/month equivalent, SAVE 17%) |
| 8.25 million points/month | $249.99/mo | 8,250,000 compute points per month. About $249.99/month billed monthly, or $2,499.99/year billed yearly ($208.33/month equivalent, SAVE 17%) |
Feature Comparison by Tier
| Feature | Free | 10 thousand points/day | 660 thousand points/month | 1.65 million points/month | 3.3 million points/month | 8.25 million points/month |
|---|---|---|---|---|---|---|
| Price (billed monthly) | $0 | ~$4.99 | ~$19.99 | ~$49.99 | ~$99.99 | ~$249.99 |
| Price (billed yearly) | $0 | $49.99/year | $199.99/year | $499.99/year | $999.99/year | $2,499.99/year |
| Monthly equivalent | $0 | $4.17 | $16.67 | $41.67 | $83.33 | $208.33 |
| Compute points | Daily allowance, lighter models only | 10,000/day | 660,000/month | 1,650,000/month | 3,300,000/month | 8,250,000/month |
| Premium models (GPT-5.5, Claude-Opus-4.7, Gemini-3.5-Flash, Nano-Banana-Pro, Veo-3.1) | — | ✓ | ✓ | ✓ | ✓ | ✓ |
| Full context length per bot (up to 2M tokens) | Limited | ✓ | ✓ | ✓ | ✓ | ✓ |
| Bot creation and revenue share | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |
| Admin console, SSO, team billing | — | — | — | — | — | — (no enterprise tier exists) |
Key Features
- One subscription, every major lab: Chat with official models from every major lab in one window without separate logins or per-provider subscriptions. Poe names GPT-5.5, Claude-Opus-4.7, Gemini-3.5-Flash, Nano-Banana-Pro and Veo-3.1 among the models a subscription reaches.
- Compute points as the single currency: Points are spent at different rates per model, so one balance covers text, image, video and audio across every provider. The entry plan's allowance resets daily; the four larger plans run on a monthly allowance.
- Bot creation and revenue share: Build a bot with no-code prompt engineering or a full API-based server bot; creators who publish are paid out of Poe's revenue-share program rather than working for exposure. Poe collects tax details from creators to facilitate those payments.
- Multimodal generation in the chat thread: Image, video and audio generation sit in the same interface as chat, so a project does not require switching tools mid-flow. Higher point allowances exist mainly because media generation consumes them quickly.
- Group chat and multi-bot threads: Add multiple bots to one thread to compare their answers side by side or chain them into a multi-step workflow. Note that in group chats, bot developers may receive limited personally identifying information about participants.
- Cross-platform sync: Native apps on web, iOS, Android, macOS and Windows keep chat history in sync without a paid sync add-on, and a subscription raises each bot's maximum input size and chat history.
Pros
- The broadest model lineup of any consumer AI chat app: one login reaches OpenAI, Anthropic, Google and Meta models alongside community-built bots, at a fraction of what four separate subscriptions cost
- A genuine creator economy: bot builders take a revenue share rather than the all-or-nothing free-fan-base arrangement most chatbot platforms offer
- One compute-point balance spans text, image, video and audio across every provider, so switching models mid-project costs nothing extra in subscriptions
- Chat history follows you between desktop and phone, and every plan costs less when billed yearly
Cons
- Your messages and uploads reach whichever outside lab or bot builder runs the bot you chat with, and builders on the API can keep them for training. Poe brokers access to models, it does not shield the conversation
- The plans have no names, only point allowances, which makes them hard to compare against a competitor's Pro or Team tier, and there is no enterprise offering: the largest plan is simply the biggest consumer allowance
- Poe publishes no real pricing page. poe.com/pricing and poe.com/plans both return 404 and the help centre sits behind a bot challenge, so the only public price surface is poe.com/subscription_plans
- Compute points add a layer of arithmetic between you and the cost: per-model rates vary, allowances are daily on the entry tier and monthly above it, and a heavy media-generation session can consume a month's balance quickly
- It is a broker, not a platform: no SOC 2 or other certification is published, there is no admin console, SSO or team billing, and support is reported as slow
Data Handling
- Training-data policy
- Read this before sending anything sensitive. Poe's privacy policy (last updated 30 April 2026, Quora, Inc. as data controller) states that "Developers that create bots and apps on Poe using APIs may view and store your chats on their servers to train their models", and that "Our third party AI model providers and third party developers may receive details about your interactions with bots and apps on Poe (including the contents of your chats and photos or documents you upload)". Poe does not state whether it itself trains on conversations. Account identity is withheld from those parties, "Your Poe user account information (e.g., username and email) is not shared with AI model providers or developers", except in group chats, where developers may receive limited personally identifiable information. The Poe policy publishes no retention period or deletion procedure of its own and defers to the Quora Privacy Policy, which it incorporates by reference, for exercising privacy rights. Poe collects email and phone number for account creation, IP address to personalise outputs (some bots receive a rough IP to answer location-specific questions), device information, and, with consent, phone contact lists. Billing is handled by a third-party provider, and creator payouts involve a third-party payment processor plus a tax service provider that receives tax ID and address.
- Compliance
- None published. Poe claims no SOC 2 · ISO or other certification and runs no public trust centre; its privacy policy incorporates the Quora Privacy Policy by reference for privacy rights.
Frequently Asked Questions
What does Poe actually cost?
Poe is free to start, and five paid plans sold by point allowance run from $4.99 a month for 10 thousand points/day to $249.99 a month for 8.25 million points/month, per poe.com/subscription_plans on 28 October 2026. Billed yearly, the same plans cost $49.99 to $2,499.99, about 17% less.
Can you use Poe without paying?
You can. Poe's free plan comes with a daily compute-point allowance, and unused free points do not roll over. Poe's plan page says paid plans send "far more messages than free users" and reach more of the top models. That page does not state the size of the free allowance.
What should you use instead of Poe?
OpenAI's native Codex agent and deeper first-party integrations point toward ChatGPT instead of a multi-vendor hub. Perplexity fits users who need cited, source-backed research answers rather than raw model access. OpenRouter is the developer-first option, offering direct API access to the same model catalog without Poe's consumer chat interface.
What separates Poe from ChatGPT?
Poe's advantage is model breadth: one subscription reaches models from OpenAI, Anthropic, Google, and Meta instead of locking you into a single vendor. ChatGPT's advantage is depth within OpenAI's own product line, including its native Codex coding agent and first-party integrations a third-party client like Poe cannot replicate.
How long does it take to get going with Poe?
Chatting starts immediately after signup: create a free account at poe.com or in the mobile app, then browse the model and bot list within the free plan's daily point limit. Upgrading to a paid tier raises that point pool and unlocks premium models, and developers can request an API key from poe.com/api/keys to build against Poe's OpenAI-compatible API.
Top Alternatives
- ChatGPT: Multi-model access from one subscription is Poe's whole point. ChatGPT counters with its native Codex agent and deeper first-party features inside OpenAI's own ecosystem.
- Perplexity: Model variety across vendors is where Poe wins. Perplexity instead built its reputation on sourced, citation-backed answers to research questions, not on stacking multiple chat models behind one login.
- OpenRouter: Poe ships a ready-made chat UI on top of the models. OpenRouter strips that away for direct, unbundled API access to the same model catalog.
Hokai guides covering Poe
- Poe AI Alternatives in 2026: OpenRouter, Raycast, Msty and When One Model Is Enough: Poe bills in compute points behind a login. OpenRouter charges list price plus a 5.5% fee, Raycast Pro is $8 a month and Msty runs models locally for free.
- How to Choose an AI Assistant: ChatGPT vs Claude vs Copilot vs Grok vs Poe: ChatGPT, Claude, Microsoft Copilot, Grok or Poe? Plans and prices read from each vendor on 27 Oct 2026, plus three questions to settle which assistant fits you.
Also mentioned in
- Best AI Chatbots in 2026: Free and Paid Picks for Everyday Use: ChatGPT, Claude, Gemini, Mistral's Vibe, Poe and Character.ai compared on price and actual job fit, not raw model rank, with a decision table and 2026 pricing.
- ChatGPT Ads Explained (2026): Which Plans Show Them and How to Avoid Them: ChatGPT ads show on Free and Go, not Plus or Pro. See what they look like, what data they use, the image-ad test and five ways to avoid them (Oct 2026).