Pollinations
Open infrastructure giving builders one API for image, video, speech, music, text, agents and MCP tools
Reach for Pollinations when you want four modalities answering prompts in one afternoon: one API, a large named model shelf, and 1 pollen = $1 pay-as-you-go so experiments stay cheap. The September 2026 additions — Claude Sonnet 5.5 with up to 1M tokens, the Anthropic-compatible /v1/messages endpoint, Gemini 3.5 transcription and ElevenLabs stem splitting — make it more than a toy endpoint. It is not a fit for work that needs an SLA or a forecastable flat invoice; keep OpenRouter or Replicate in mind if you only want bare inference, and treat Pollinations as the layer that also hands you agents and MCP tools.
Verified 53m ago · liveness 86/100 · cite: rightaichoice.com/tools/pollinations
- Developers shipping a multimodal prototype fast
- Apps that want users to pay their own usage via BYOP
- Builders swapping between many image, text, audio and video models
- Agent developers wiring hosted MCP tools like Exa Search or FFmpeg
- Production workloads that need an SLA or contractual uptime
- Teams requiring formal vendor onboarding and a dedicated support tier
- Finance teams needing a flat, forecastable monthly invoice
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip Pollinations if you need a contracted SLA, a dedicated support tier or a flat forecastable monthly invoice, or if your workload must run on private infrastructure in a specific region.
Everything meters back to pollen at 1 pollen = $1, so a chatty agent that fans out to several models per request multiplies your spend without a per-run cap
Pay-as-you-go pollen at 1 pollen = $1 suits solo builders and small teams running spiky experiments who would rather not commit to a monthly seat. Teams that need a forecastable invoice or vendor SLA are better served by contracted inference providers; teams that only need bare model access can look at OpenRouter or Replicate and skip the agent and MCP layer entirely.
In short
Pollinations — Open infrastructure giving builders one API for image, video, speech, music, text, agents and MCP tools. Best for Developers shipping a multimodal prototype fast, Apps that want users to pay their own usage via BYOP, Builders swapping between many image, text, audio and video models. Free to start; paid plans from $1.
No head-to-head comparison for Pollinations yet — the alternatives below are the closest matches we list.
What's new in Pollinations
Checked todayAcross the latest 5 updates: 5 feature updates.
Claude Sonnet 5.5
anthropic/claude-sonnet-5.5 handles chats with images, tools, adaptive reasoning and up to 1M tokens of context.
Claude Code, meet Pollinations
A new Anthropic-compatible /v1/messages endpoint supports streaming, tools and thinking across chat-capable models; point ANTHROPIC_BASE_URL at gen.pollinations.ai.
Lightning Image Turbo
Generate images from a prompt with up to two reference images for guidance through the image API.
Audio gets more useful
Transcribe with Gemini 3.5 for word timestamps and speaker labels, or split a recording into two or six stems with ElevenLabs.
More voices for speech
openai/tts-1 and openai/tts-1-hd are now available through the existing speech endpoints.
What people actually say about Pollinations — is it worth it?
We ran a structured research pass across product reviews, community discussions, and post-purchase forum threads to surface the patterns vendors won't publish themselves. Below: the recurring strengths, the hidden costs people mention most, and the cohort that consistently regrets adopting this tool.
12 mentions across 4 sources (Hacker News, YouTube, GitHub, Lemmy), 26 more we could not attribute · researched Sep 25, 2026.
Weighted by the 38 posts each of 4 sources contributed.
- +No signup, no API key, no credit card — send a prompt, get output in seconds.
- +Single REST endpoint covers text, image, audio, and video generation.
- +Reported faster than Hugging Face for lightweight LLM backend work.
- +OpenAI-compatible endpoint means existing clients can point at it with minimal changes.
- +5,127 GitHub stars and a live ecosystem of plugins, tools, and daily health-checkers.
- −Documented --stream flag on gen text was never registered and fails hard.
- −--json gen text --output writes the file but prints nothing to stdout.
- −gen audio --format flac still defaults to writing speech.mp3 — extension lies.
- −polli gen audio --play hangs indefinitely on Windows when file isn't decodable.
- −No SLAs, no content moderation, and limited support for production use.
- • Pollen wallet creates a soft dependency — heavy usage may push you toward earning or buying Pollen rather than staying purely free
- • No support contract means outages cost you engineering time instead of being covered by an SLA
- • Production monetization via user spending may carry implied revenue-sharing or platform terms not spelled out in the data
Viability Score
How well maintained and how widely used is Pollinations? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this
Last calculated: October 2026
How we score →Key Features
- One REST API for image, video, speech, music and text generation
- 46 image models including FLUX.3 Image Klein and Qwen Image 2.1 Turbo
- 116 text models including DeepSeek V4.1 Flash and Gemini 4 Argon
- 41 audio models, 26 video and 3D models, 5 embeddings models
- Claude Sonnet 5.5 with images, tools, adaptive reasoning and up to 1M token context
- Anthropic-compatible /v1/messages endpoint with streaming, tools and thinking
- OpenAI-compatible endpoints for drop-in client swaps
- Lightning Image Turbo generation with up to two reference images
- Gemini 3.5 transcription with word timestamps and speaker labels
- ElevenLabs stem splitting into two or six stems
- Speech through openai/tts-1 and openai/tts-1-hd on existing speech endpoints
- 48 hosted agents callable like models, plus user-published prompt and code agents
- Hosted MCP servers: FFmpeg, Exa Search, connectors, private computer with web search
- Media storage returning a link for image, audio and video model calls
- Per-request safety checks for personal data, secrets, NSFW and prompt attacks
About Pollinations
Pollinations is open infrastructure aimed at builders rather than end users: one API that generates images, video, speech, music and text, with hosted agents, MCP tools and media storage sitting on the same account. The homepage currently counts 46 image models, 116 text models, 41 audio models, 26 video and 3D models and 5 embeddings models, plus a community shelf of 104, with recent arrivals like FLUX.3 Image Klein, Qwen Image 2.1 Turbo, DeepSeek V4.1 Flash, Gemini 4 Argon, MAI Transcribe 2 and Mirelo SFX 2.0 listed by name. The September 2026 changelog is the most useful thing to read before you commit code: Claude Sonnet 5.5 landed with images, tools, adaptive reasoning and up to 1M tokens of context, and a new Anthropic-compatible /v1/messages endpoint (point ANTHROPIC_BASE_URL at gen.pollinations.ai) covers streaming, tools and thinking, so Claude Code and similar clients can talk to it. Audio got transcribe-with-word-timestamps and speaker labels via Gemini 3.5, ElevenLabs stem splitting into two or six stems, and openai/tts-1 and tts-1-hd on the existing speech endpoints. Beyond inference, 48 agents can be called like models and you can host your own prompt or code agents; hosted MCP servers cover FFmpeg, Exa Search, connectors and a private computer with web search and app access. Media storage turns uploads into links for model calls, and per-request safety checks strip personal data and keys or block unsafe prompts before they reach a model. Billing is pay-as-you-go on pollen at 1 pollen = $1, topped up via Stripe, with free Quest pollen for solving GitHub Quests or trying models, and BYOP lets your own users spend their own balance inside your app. It overlaps with OpenRouter and Replicate but bundles agents, MCP tools and app hosting alongside inference.
Behind the Verdict
The appeal here is breadth without setup friction. One account gets you image, text, audio, video and 3D generation plus 5 embeddings models, and the catalog is genuinely current — Claude Sonnet 5.5, FLUX.3 Image Klein, Qwen Image 2.1 Turbo, DeepSeek V4.1 Flash, Gemini 4 Argon, MAI Transcribe 2, Mirelo SFX 2.0 and Kandinsky 6.0 Lite all appear by name on the homepage or in the changelog. What separates it from a plain inference aggregator is the surrounding plumbing. 48 hosted agents are callable like models, and you can publish your own prompt or code agents; hosted MCP servers for FFmpeg, Exa Search, connectors and a private computer mean an agent can transcode a file or run a search without you standing up infrastructure. The Anthropic-compatible /v1/messages endpoint is the quiet highlight — pointing ANTHROPIC_BASE_URL at gen.pollinations.ai gives Claude Code a working backend with streaming, tools and thinking, and the OpenAI-compatible endpoints do the same for other clients. Audio side, Gemini 3.5 gives you word timestamps and speaker labels for transcription, ElevenLabs splits a recording into two or six stems, and openai/tts-1 and tts-1-hd are reachable through the existing speech endpoints. Weaknesses are real. The model shelf moves month to month, so a pipeline pinned to one version can shift under you — pin what you depend on. Support is community-shaped (Discord, GitHub, open in the open) rather than contract-based, and the vendor page states no SLA. Publishing models, agents and apps is in alpha and needs publisher access, so the earnings side is not a finished product yet. Billing is metered pollen rather than a flat subscription, which is great for spiky experimentation and awkward for a finance team that wants a predictable line item. Treat it as the prototyping ground and the agent-tooling layer, and move client-facing deliverables onto contracted infrastructure.
Researching Pollinations? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Real-world workflow fit
Concrete scenarios for the personas Pollinations actually fits — and what changes day-one when you adopt it.
You run npx @pollinations/cli, grab an API key, and hit the image endpoint with Lightning Image Turbo to render thumbnails from two reference images, then switch the same call to a text model to caption them.
Outcome: A working multi-modal demo in an afternoon with no separate contracts per modality and one pollen balance covering all calls.
You turn on BYOP so visitors sign in with Pollinations and set their own spending limit, while your app calls the OpenAI-compatible endpoints for chat and image generation.
Outcome: Your users fund their own usage instead of you absorbing inference cost, and you skip building a billing and top-up flow.
You build a code agent that calls Exa Search for research, FFmpeg to transcode the resulting media, and Claude Sonnet 5.5 through the Anthropic-compatible /v1/messages endpoint for reasoning.
Outcome: A hosted multi-step agent that runs without you standing up separate search and media infrastructure.
Use Cases
- Prototype a multi-modal app over a weekend without signing up for a separate model provider
- Generate an image from a prompt with up to two reference images via Lightning Image Turbo
- Transcribe a recording with word timestamps and speaker labels using Gemini 3.5
- Split a track into two or six stems with ElevenLabs for remix or practice
- Point Claude Code at the Anthropic-compatible endpoint and run batch generation from the terminal
- Let your users sign in and spend their own pollen via BYOP instead of your credits
- Chain an agent with the Exa Search or FFmpeg MCP server to research then transcode in one flow
- Teach generative AI concepts with image, text, audio and video from one endpoint
Models Under the Hood
as of 2026-10-21
Limitations
- No SLA or uptime guarantee is described on the site.
- The model lineup moves month to month — the September 2026 changelog alone swapped in Claude Sonnet 5.5, Lightning Image Turbo, Gemini 3.5 transcription, ElevenLabs stem splitting and openai/tts-1 and tts-1-hd — so a pipeline pinned to one model version can shift under you.
- Support is community-oriented via Discord and GitHub rather than contract-based.
- Publishing models, agents and apps is in alpha and needs publisher access.
- Free Quest pollen comes with usage limits that are not spelled out.
- Billing is metered pollen rather than a flat plan, so spend tracks your request volume.
as of 2026-10-22
Verification history
We have re-verified Pollinations 24 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-checked, vendor evidence unchanged
- — re-checked, vendor evidence unchanged
Showing the 6 most recent of 24 verification passes.
Free to cite with attribution — this page re-verifies continuously.
12-month cost
Project the real annual outlay, including the implied monthly cost when only an annual tier is published.
Vendor list price only. Add-on usage, seat overages, and contract minimums are surfaced under Hidden costs & gotchas.
Plans compared
For each published Pollinations tier: who it actually fits, and what it adds vs. the previous tier. Cross-reference the cost calculator above for projected annual outlay.
Free Quest pollen
$0
Ideal for
Developers who want to try models or ship a first app or agent without paying, and are willing to solve GitHub Quests to earn credit.
What this tier adds
Free entry point: pollen earned by solving GitHub Quests, trying models, or building an app or agent, with usable limits not published.
Pay as you go
1 pollen = $1
Ideal for
Builders running regular generation across several modalities who want metered spend without a monthly subscription.
What this tier adds
Adds paid top-ups at 1 pollen = $1 via Stripe, covering every model, agent and tool, on top of any free Quest pollen.
Where the pricing makes sense
The company stage and team size where Pollinations's pricing actually pencils out — and where peers do it cheaper.
Pay-as-you-go pollen at 1 pollen = $1 suits solo builders and small teams running spiky experiments who would rather not commit to a monthly seat. Teams that need a forecastable invoice or vendor SLA are better served by contracted inference providers; teams that only need bare model access can look at OpenRouter or Replicate and skip the agent and MCP layer entirely.
Setup time & first value
How long it actually takes to get something useful out of Pollinations — broken out by persona, not the marketing-page minute.
Solo builder: an API key and a first image call in minutes, since keys come from the dashboard and the CLI installs with one npx command. App builder adding BYOP: an hour or two, mostly wiring your own sign-in flow to the spending limit. Agent developer: half a day, since you need to pick MCP servers and test streaming and tool calls through the Anthropic-compatible endpoint.
Switching to or from Pollinations
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From OpenRouter: point your OpenAI-compatible client at the Pollinations endpoints and reuse existing model IDs where the catalog matches
- →From Replicate: move image and video calls to the unified endpoint and swap per-model auth for a single pollen balance
- →From a direct Anthropic setup: set ANTHROPIC_BASE_URL to gen.pollinations.ai and keep your Claude Code workflow with streaming, tools and thinking intact
- ↗To OpenRouter: keep the OpenAI-compatible call shape and swap base URL and key, dropping agents and MCP tools you relied on
- ↗To a contracted inference provider: re-point model IDs, add per-modality contracts and rebuild any agent orchestration you had on hosted MCP servers
Integrations
Resources & Guides
- Documentationpollinations.ai
pollinations.ai
Build AI apps with easy APIs, free compute, and community support
- Resourcepollinations.ai
pollinations.ai
Build AI apps with easy APIs, free compute, and community support
- Resourcepollinations.ai
pollinations.ai
Build AI apps with easy APIs, free compute, and community support
Tutorials & Learning
Unleash Your Creativity with Pollinations.AI: The AI-Powered Platform that Transforms Your Ideas int
phill ai
Pollinations AI 無料API 無制限の無料Claude、GPT、Gemini、画像、動画、音声API（制限なし）
Coding Shiksha
YouTube returned 6 videos for “Pollinations”, and we withheld 4: 4 could not be judged, because “Pollinations” is a single word that other videos use for other things. Showing the 2 we can prove are about Pollinations.
Official links
Tools that pair well with Pollinations
Common stack mates teams adopt alongside Pollinations, with the specific reason each pairing earns its keep.
Invideo AI
Agentic AI video editor: one plain-language instruction lands across every shot on a real multitrack timeline.
Envato Elements
Envato Elements is a flat-fee subscription bundling 29.9M+ stock assets with in-subscription AI video, image, voice, music and audio tools under a lifetime
Luma AI Genie
Luma AI Genie: AI video generation with Ray 3.2 frame-level control and brand-consistent agents
Alternatives to Pollinations
View all
Invideo AI
Agentic AI video editor: one plain-language instruction lands across every shot on a real multitrack timeline.
Envato Elements
Envato Elements is a flat-fee subscription bundling 29.9M+ stock assets with in-subscription AI video, image, voice, music and audio tools under a lifetime
Luma AI Genie
Luma AI Genie: AI video generation with Ray 3.2 frame-level control and brand-consistent agents
Frequently Asked Questions
Categories
Best-of guides
Used Pollinations? Help shape our editorial sentiment research.