What is TinyFish?
Built for AI agents that need current and reliable web data, TinyFish combines live search, page fetching, browser sessions, and multi-step web automation in one platform. It can search fresh web sources, render dynamic pages, and return clean content as structured JSON, Markdown, or HTML.
The platform can navigate websites, complete forms, access authenticated portals, and extract structured results from complex workflows. Its browser infrastructure supports persistent sessions, anti-bot handling, credential vault integration, and connections with MCP-compatible clients and custom agent stacks.
Features
- Live Search API: Returns fresh, structured JSON results from live, browser-rendered web sources rather than cached indexes.
- Rendered Content Fetching: Converts any rendered URL into clean, token-efficient Markdown, JSON, or HTML.
- Multi-Step Web Agent: Navigates pages, fills forms, authenticates into websites, and extracts structured results from production workflows.
- Cloud Browser Sessions: Provides persistent browser state, dynamic rendering, engine-level stealth, and sub-250ms cold starts.
- Unified Integration: Combines Search, Fetch, Agent, and Browser products under one API key and credit pool, with SDK, CLI, and MCP support.
Use Cases
- Monitor competitor prices across dynamic product catalogs.
- Track breaking news, earnings updates, and other time-sensitive web data.
- Automate insurance quote collection and premium comparisons.
- Collect real-time hotel availability and inventory information.
- Analyze social listening data and crowd-sourced sentiment.
- Access authenticated vendor portals and compliance dashboards.
- Run autonomous quality assurance tests on changing websites.
- Build research pipelines that transform web pages into structured data.
How It Works
Get an API Key
Sign up to receive one key for TinyFish Search, Fetch, Agent, and Browser products.
Connect an Agent Stack
Integrate through an MCP-compatible client, SDK, CLI, Python, npm, or a direct custom integration.
Run a Web Operation
Search the live web, fetch clean content from a URL, start a browser session, or deploy an agent for a multi-step task.
Receive Structured Results
Use returned JSON, Markdown, or HTML in an AI application, automation workflow, or downstream data pipeline.
FAQs
-
What does stealth browser technology mean in TinyFish?
TinyFish uses C++ patches compiled directly into Chromium rather than JavaScript injection. Its browser achieved an 85% pass rate across services including Akamai, Cloudflare, DataDome, PerimeterX, and Imperva. -
How accurate is the TinyFish Web Agent?
The Web Agent achieved 89.9% overall accuracy on the Mind2Web benchmark and 81.9% on hard tasks. TinyFish states that all 300 benchmark runs are open-sourced and independently verifiable. -
Can TinyFish agents log into websites?
Yes. Vault integration allows agents to authenticate into portals, navigate gated dashboards, and extract content without exposing credentials to the language model. -
Are failed operations charged?
No. TinyFish only charges for successful operations, while failed runs cost nothing. -
Do monthly credits roll over?
No. Starter and Pro credits reset at each billing cycle. Pay-as-you-go usage has no billing-cycle credit balance because each credit is purchased as it is used.