MIAPI

Web-grounded AI answers API with inline citations and OpenAI compatibility

Advanced API
Screenshot of MIAPI, Web-grounded AI answers API with inline citations and OpenAI compatibility

What is MIAPI?

MIAPI is an OpenAI-compatible answers API that pairs real-time web search with cited, confidence-scored responses, plus knowledge mode, news and image search, streaming and a Python SDK. Query packs run from $9 for 2,500 queries, with a free tier of 100 queries per month.

MIAPI is a developer-facing answers API that runs a live web search behind every request and returns a synthesized response with inline citation markers, a source list and a confidence score. The core endpoint accepts a question of 2 to 1000 characters and handles the retrieval, synthesis and formatting in one call. Typical response time is quoted at roughly 1.1 to 1.5 seconds, and repeat queries are served from a cache that can return in about a millisecond. The design assumes you already have a codebase. A dedicated chat completions endpoint mirrors OpenAI's format, so existing integrations can be repointed by swapping the base URL and the model name, while the Python SDK covers sync, async and streaming usage. Beyond the main answer mode there are search-only responses for teams that want raw results and their own model, a knowledge mode that takes up to 10K characters of your own documents and blends them with or replaces the web search, plus separate news and image endpoints. Filters let you restrict or exclude domains, force a response language, set a timezone and choose text, short, JSON or markdown output. A notable extra is a kids style, which returns short spoken answers, handles counting and spelling questions, and can block topics such as politics or death behind a parental flag and a custom reply. Conversation memory arrives either through a conversation id that retains 30 exchanges for 24 hours of idleness, or through context you send yourself. This fits teams building assistants, research tools, RAG pipelines or news monitors who need attributable answers without assembling a search stack themselves. It competes on price and feature bundling against Perplexity, Tavily, Exa and Google grounding APIs rather than on model breadth, and the published comparison puts it at $2.50 to $3.60 per 1,000 queries depending on pack size.

How do you use MIAPI?

  1. 1Get an API key
    Create an account on the signup page and generate a key. The free tier activates 100 queries per month with a single key, enough to test every endpoint.
  2. 2Test the question in the playground
    Use the on-page playground to send a demo question without signing in. Toggle citations and sources to see the response shape before writing any code.
  3. 3Send your first answer request
    Post a question to the answer endpoint with a bearer token. Set citations and include_sources to true, then read the answer, sources array and confidence value from the JSON response.
  4. 4Repoint an existing OpenAI client
    If you already call chat completions, change the base URL to the MIAPI endpoint and set the model to miapi-grounded. The rest of your request and response handling stays the same.
  5. 5Add private documents or raw search
    Switch the mode parameter to knowledge and pass your own text, or to search for unprocessed results to feed your own model or retrieval pipeline.
  6. 6Wire it into an MCP client
    Download the MCP server file, install the MCP tooling and set your key so assistants such as Cursor or Claude Desktop can call web search with citations directly.

Pros and cons

Pros

  • Drop-in OpenAI compatibility means existing chat completions code works after changing the base URL and model nameAI
  • Every answer can include [1][2] citation markers linked to source URLs, which makes outputs verifiableAI
  • Feature set bundles knowledge mode, news search, image search and raw search at one price pointAI
  • Query packs never expire, so low-volume projects are not billed for idle monthsAI
  • Free tier of 100 queries per month plus a no-auth playground endpoint for testing before signupAI

Cons

  • Smaller, newer vendor than the incumbents it benchmarks against, with little independent track record to checkAI
  • Free tier is capped at 60 requests per minute and the public demo endpoint at 20 per minute per IPAI
  • Pay-as-you-go packs are sold up front rather than metered monthly, so unused volume is capital sitting idleAI
  • Knowledge context is limited to 10K characters per call, which rules out large document sets without chunkingAI
  • No official SDK listed beyond Python, so other language users work against raw HTTPAI

How much does MIAPI cost?

Free trial

A permanent free tier provides 100 queries per month with one API key and access to all endpoints, refilled monthly. A no-auth playground endpoint allows demo queries before any account exists.

Learn more

Pricing

Free tier at $0 with 100 queries per month, 1 key and 60 requests per minute. Paid packs are one-time purchases that never expire: Starter $9 for 2,500 queries, Developer $33 for 10,000, Pro $90 for 30,000 and Scale $250 for 100,000 ($2.50 per 1,000). Enterprise pricing is custom with volume discounts and an on-premise option.

Learn more

Support

Contact is handled by a form on the site or by email to support@miapi.uk, which is also listed for partnership and sales inquiries.

Integrations

An OpenAI-compatible chat completions endpoint, a Python SDK with sync and async support, and an MCP server for assistants such as Cursor, Claude Desktop and Windsurf. Raw search output is aimed at LangChain and RAG pipelines.

Learn more

Features

Grounded answers with inline citations, sources and confidence scores; conversation memory via conversation id or supplied context; knowledge mode for your own text; search-only, news and image endpoints; SSE streaming; domain include and exclude filters; language, timezone and output format controls; a kids answer style with blockable topics; usage and key management endpoints.

Frequently asked questions about MIAPI

  • How much does MIAPI cost?
    Free tier at $0 with 100 queries per month, 1 key and 60 requests per minute. Paid packs are one-time purchases that never expire: Starter $9 for 2,500 queries, Developer $33 for 10,000, Pro $90 for 30,000 and Scale $250 for 100,000 ($2.50 per 1,000). Enterprise pricing is custom with volume discounts and an on-premise option.
  • Does MIAPI offer a free trial?
    A permanent free tier provides 100 queries per month with one API key and access to all endpoints, refilled monthly. A no-auth playground endpoint allows demo queries before any account exists.
  • Does MIAPI have an API?
    Yes, MIAPI offers an API.
  • How do you use MIAPI?
    The walkthrough on this page covers 6 steps: 1. Get an API key 2. Test the question in the playground 3. Send your first answer request 4. Repoint an existing OpenAI client 5. Add private documents or raw search 6. Wire it into an MCP client.
  • What platforms does MIAPI support?
    MIAPI is available on Web App.
  • What does MIAPI integrate with?
    An OpenAI-compatible chat completions endpoint, a Python SDK with sync and async support, and an MCP server for assistants such as Cursor, Claude Desktop and Windsurf. Raw search output is aimed at LangChain and RAG pipelines.

Status

StatusActive
Views0
Outbound clicks0
Added10/21/2026

Platforms

Web App

Pricing

FreePaid