Starting today, TinyFish Web Search and Fetch are free.
Free for every developer. Free for every agent. Free across every surface you already build on — the REST API, the MCP server, Python and TypeScript SDKs, TinyFish CLI, Skill, and every integration: OpenClaw, Hermes Agent, n8n, Dify, LangChain, CrewAI, and more.
One sign-up at agent.tinyfish.ai, one key, that’s it!
Search is the entry point to the web; period. Every question an agent answers, every fact it verifies, every page it reads starts with a Search.
And once an agent has the URL, it needs to actually read the page — clean, structured, agent-ready. That's Fetch. they're the two primitives every agent needs to use the web.
Together, they are the two most fundamental things agents, and even humans, do on the web. And fundamental things shouldn't cost you money. That's why we made Search and Fetch free.
Both endpoints run on TinyFish's own browser infrastructure — a custom Chromium fleet with stealth handling, parallel execution, and full JavaScript rendering built in. Owning the stack end-to-end is what lets us make Search and Fetch both free and fast. There’s no middlemen, markup, or tradeoff.
Web search built for agents. Structured JSON results, low latency, rank-stable across calls. Use it as a discovery step in front of Fetch, the retrieval layer in a RAG pipeline, or a quick verification step inside a multi-turn agent run. Free just the like the web was meant to be.
Most web fetches dump raw HTML into your agent's context, even nav bars, scripts, cookie banners, all of it. Your model pays for every token of that noise.
TinyFish Fetch cleans it up. Hand it any URL, and we render the page in a real, full browser with JavaScript, SPAs, and dynamic content, strip out everything unnecessary, and return clean Markdown, JSON, or HTML. Whichever your agent needs.
Failed URLs don't count against your quota. And, it’s still free.
We're not just giving you free web access. We're cutting your token costs on every single call.
REST API
api.search.tinyfish.ai and api.fetch.tinyfish.ai. Grab your API key from agent.tinyfish.ai, that’s it!
curl "<https://api.search.tinyfish.ai?query=web+automation+tools>" \\
-H "X-API-Key: $TINYFISH_API_KEY"curl -X POST <https://api.fetch.tinyfish.ai> \\
-H "X-API-Key: $TINYFISH_API_KEY" \\
-H "Content-Type: application/json" \\
-d '{"urls": ["<https://www.tinyfish.ai/>"]}'MCP
Drop the server URL into Claude, Cursor, Codex, ChatGPT desktop, or any MCP-aware client.
// MCP — drop into Claude, Cursor, Codex, or any MCP-aware client
{
"mcpServers": {
"tinyfish": { "url": "<https://agent.tinyfish.ai/mcp>" }
}
}CLI
npm install -g @tiny-fish/cli. The CLI writes results to the filesystem instead of piping them through the model's context window — tokens stay low, output stays structured.
npm install -g @tiny-fish/cli
tinyfish auth loginSkill
One-line install. Works with Claude Code, Codex, Cursor, OpenCode, Antigravity, and other agents. The Skill teaches your agent when to reach for search and fetch and how to call them.
Download it now from our open-source cookbook
Python SDK pip install tinyfish
TypeScript SDK npm install @tiny-fish/sdk
Agent harnesses
Claude Code, Codex, Cursor, OpenClaw, Hermes Agent, OpenCode, Cline, Goose, Antigravity, and more.
Platform integrations n8n, Dify Marketplace, Vercel Skills on skills.sh. ChatGPT App, MCP App on Claude, more coming soon!
If your stack is in that list, you already have access. If it isn't, the REST API, SDKs, and CLI run anywhere your code runs.
Three steps.
1. Sign up at agent.tinyfish.ai and grab your API key.
2. Pick your surface.
3. Build.
The free tier includes 5 queries per minute for Search and 25 URLs per minute for Fetch. If you need higher rate limits, head to tinyfish.ai/pricing to upgrade.
Sign up free → agent.tinyfish.ai
Docs → docs.tinyfish.ai
Open source Cookbook → github.com/tinyfish-io/tinyfish-cookbook
Tag @Tiny_Fish when you ship something — we read everything, and the best builds get featured.
No credit card. No setup. Run your first operation in under a minute.
