Pollinations

What is Pollinations?

Pollinations is a unified generative AI API at gen.pollinations.ai that lets you create images, text, speech, video, 3D models and embeddings from one endpoint. A simple GET request such as an image URL with your prompt returns a picture, while OpenAI-compatible chat, image, audio and realtime endpoints cover more advanced apps. The model catalogue mixes providers, with examples in the docs including Flux, Z-Image, GPT Image, Claude, Gemini, DeepSeek, Veo, Seedance, ElevenLabs and Whisper. A web Play page lets non-developers try image, video and audio generation in the browser.

Pollinations is built by a Berlin-based team and released under the MIT licence, with its code, roadmap and decisions published on GitHub, where the main repository has about 5,000 stars. The project says it powers more than 500 community projects and lists apps, Discord, Telegram and WhatsApp bots built on its API. Billing runs through Pollen credits at roughly $1 each, plus free Pollen earned through Quests. The platform is actively developed, with new models and a redesigned Play page announced on 10 October 2026.

Key Features

• Unified generation API: Pollinations serves text, image, video, audio, 3D and embeddings from a single endpoint at gen.pollinations.ai. You manage one key and one billing balance instead of separate accounts with each model provider.
• URL-based image generation: Pollinations returns an image from a plain URL such as /image/ followed by your prompt. Width, height, seed and model are set as URL parameters, so you can embed generated images in a web page or bot with no backend code.
• OpenAI and Anthropic compatibility: Pollinations exposes OpenAI-compatible chat, responses, images, speech and embeddings endpoints, plus Anthropic Messages support. Existing OpenAI SDK, Vercel AI SDK or Claude Code setups can switch over by changing the base URL and key.
• Wide model catalogue: The catalogue spans text models from OpenAI, Anthropic, Google and DeepSeek plus image models such as Flux.2 Pro. Video models include Veo 3.1 Fast and Seedance 2.0. Live model lists and prices come from the /models endpoints.
• Audio and realtime: Pollinations offers text-to-speech, transcription, voice changing and voice isolation endpoints. A WebSocket realtime API supports streaming voice conversations using OpenAI-compatible message formats, which suits voice assistants and live chat features.
• User wallets and developer earnings: Connect User Wallets lets your app’s users sign in and pay for their own generations, so the developer pays nothing for usage. Developers can opt in to a 25% markup on base rates, which is credited to their balance.
• Community models and agents: Community members can register their own OpenAI-compatible model endpoints with up to three fallback models. Managed agents turn a system prompt, base model and optional tools into a reusable model ID called through the same API.
• CLI and MCP server: The @pollinations/cli package and an MCP server let AI assistants and coding tools call Pollinations directly. The CLI can also reconfigure coding harnesses such as OpenCode and Pi to run on Pollinations models.

Pros

• Genuinely open source: Pollinations publishes its full code under the MIT licence, along with its roadmap and decisions. You can inspect how requests are routed, report issues on GitHub and avoid lock-in to a closed vendor.
• Very low barrier to start: Pollinations can generate an image from a single URL in the browser or terminal. Switchtools highlights this near-zero setup as its biggest strength for students, hobbyists and quick prototypes.
• Free route available: Free models and Quest Pollen mean you can build and test without paying, and the project says no credit card is required. Paid Pollen at about $1 each only becomes necessary for premium models.
• Many models, one bill: One account reaches text, image, video and audio models from several major providers. Comparing Flux against GPT Image, or Claude against Gemini, takes a parameter change rather than a new contract.
• Apps can pay for themselves: The user wallet system shifts generation costs to each end user, with an optional 25% developer markup. Indie developers can ship free-to-try AI apps without carrying the bill for every visitor.
• Fast-moving development: The project ships new models and features almost daily, according to the news feed in its GitHub README. Recent additions include MAI voices, PDF input for Claude and GPT, and a new Play page.

Cons

• No formal SLA: Pollinations offers no formal service level agreement or enterprise support, according to ToolJunction. Production apps that need guaranteed uptime should plan fallbacks, because downtime on the API halts any app that depends on it.
• Free tier can be slow: Directory reviewers report that free and anonymous requests can fail or slow down at peak times. The docs limit legacy public keys to 1 Pollen per IP per hour, which suits testing rather than real traffic.
• Premium models cost Pollen: Models marked paid-only cannot be paid for with free Quest Pollen. The strongest text, image and video models therefore need a paid balance, and per-model rates vary widely.
• Changing catalogue and docs: Model IDs, tiers and defaults change often, and older directory pages still describe Seed, Flower and Nectar tiers that the current docs no longer mention. Check the live /models endpoint before hard-coding a model name.
• Public media links: Uploaded media URLs are publicly readable by anyone with the link and are kept for 30 days. Avoid sending private or client files through the upload service without considering who could access them.
• Licensing depends on model: Commercial use depends on each underlying model’s own licence, as Switchtools points out. Businesses must check the terms for every model they use rather than relying on Pollinations’ MIT licence alone.

Who is Using Pollinations?

• App and bot developers: Add image, text and voice generation to web apps and Discord or Telegram bots.
• Students and educators: Learn prompt and API basics without a paid account.
• Indie makers: Ship AI apps where users pay through their own Pollen wallets.
• AI agent builders: Connect assistants through the MCP server and CLI.

Pricing

Pollinations runs on pay-as-you-go Pollen credits at roughly $1 per Pollen, with free models and Quest Pollen for no-cost use (official docs, October 2026).

Older directory listings still describe anonymous access and Seed, Flower and Nectar tiers, which no longer appear in the official documentation. Model rates move as providers change their prices, so read the live model list before estimating a monthly budget.

What Makes Pollinations Unique?

An open-source, community-built API that pays developers is a rare mix, and it defines Pollinations. Most multi-model gateways are closed commercial services, whereas Pollinations publishes its code under MIT, lets anyone register their own model with fallbacks, and lets app users pay through their own wallets with an optional developer markup. The URL-based image endpoint also keeps the barrier to entry lower than almost any rival. Free Quest Pollen rewards people who contribute to the project rather than only those who pay.

Compatibilities and Integrations

• APIs: REST endpoints, OpenAI-compatible routes, Anthropic Messages support and a realtime WebSocket.
• SDKs and tools: Works with the OpenAI SDK, Vercel AI SDK, the Pollinations CLI and an MCP server.
• Bots and apps: Community integrations for Discord, Telegram and WhatsApp, plus 30+ web and mobile apps.
• Source code: MIT-licensed repository on GitHub for self-study and contributions.

Pollinations Tutorials

A first image request takes under a minute, and a full API key setup only a little longer.

1. Open pollinations.ai/play to test image, video or audio prompts in the browser.
2. Sign in at enter.pollinations.ai and create a secret sk_ key for server use.
3. Call gen.pollinations.ai/image/ with your prompt and model as URL parameters.
4. Point an OpenAI SDK at gen.pollinations.ai/v1 to use chat and speech models.
5. Check the /models endpoint for prices, then top up Pollen if you need paid models.

How We Rated It

Summary

Pollinations suits developers, students and indie makers who want one open-source API for many generative models without upfront cost. Its main limitation is reliability, since there is no SLA and free requests can stall at busy times. Paid-only models and shifting model IDs also need some attention. Start with the Play page and a free model, keep secret keys on your server, and read the live /models list before you commit a production app to a particular model.