ContextWire

Free search API for AI agents — 105 engines, 22 profiles, 94.3% SimpleQA accuracy, MCP server with 5 tools

造訪 contextwire.dev

⚡ 94.3% on SimpleQA benchmark — best in class

The context API that actually answers questions

Multi-source research engine for AI apps. Google, Bing, Wikipedia, CrossRef, and 100+ engines — unified in one API with LLM extraction. Zero cost per query.

Try It Free → Read the Docs

94.3%

SimpleQA Accuracy

<4s

Avg Response

100+

Search Engines

$0

LLM Cost

Built for AI developers

Everything your AI app needs to answer questions accurately, backed by real sources.

🔍

Multi-Engine Search

Google + Bing via SearXNG, with automatic retry and result deduplication. 22 search profiles for web, news, academic, code, social, and more.

100+ engines

📖

Deep Page Reading

Fetches and reads full web pages when snippets aren't enough. Extracts relevant paragraphs with table support. Clean markdown or text output.

Markdown + text

🧠

LLM Answer Extraction

Trinity model extracts precise answers from context. Zero cost — runs on OpenRouter free tier. No API key management needed.

$0 per query

🎓

Academic Search

CrossRef, arXiv, Semantic Scholar, PubMed, and OpenAlex integration. DOIs, citations, abstracts, and metadata for any paper.

5 academic sources

🎮

Game & Media Wikis

Direct Fandom and wiki.gg API access — Minecraft, Dota 2, Dragon Age, Rick & Morty, and more. Structured data for any franchise.

40+ wikis

📊

Wikipedia Deep Parse

Direct Wikipedia API with table extraction, infobox parsing, and Wikidata entity resolution. Structured knowledge at your fingertips.

Structured data

Simple API, powerful results

One endpoint. JSON response. Full research loop.

curl JavaScript Python

Copy

Ask a question — get an answer with sources

curl "https://contextwire.dev/api/ask?q=Who+won+the+2024+Nobel+Prize+in+Physics"
-H "Authorization: Bearer YOUR_API_KEY"

Response:

{ "answer": "John Hopfield and Geoffrey Hinton", "source": "snippets", "latency_ms": 3200, "sources": [{ "title": "Nobel Prize", "url": "..." }], "pageExcerpts": [{ "title": "...", "text": "..." }] }

Products built on ContextWire

Ready-to-use tools powered by our API — see what's possible.

💬

AskAndy

AI answer engine. Ask any factual question, get a sourced answer in seconds. Full research loop with multi-phase verification.

Try AskAndy →

✅

Fact Checker

Verify any claim instantly. Enter a statement → get TRUE/FALSE/MIXED verdict with confidence score and source citations.

Try Fact Checker →

📚

Deep Research

Generate comprehensive research reports. Combines web, academic papers, and page extraction into structured reports with citations.

Try Deep Research →

🔌

MCP Server

Add ContextWire to Claude Code or Cursor with one command. 5 tools: ask, search, extract, research, batch search.

Setup Guide →

🧪

API Playground

Interactive API tester. Try every endpoint with real data, see full request/response, copy code snippets.

Open Playground →

📊

Status Dashboard

Real-time health monitoring. Engine status, response times, uptime tracking. Full transparency.

View Status →

🔌 Works with Claude Code & Cursor

Add ContextWire as an MCP server. One command, instant access to web search, page reading, and research inside your AI coding environment.

Copy

Claude Code

claude mcp add contextwire
--transport http
https://mcp.kept.live/mcp
-- --header "Authorization: Bearer YOUR_KEY"

Try it now

No signup needed — test the API right here.

Ask

Bring Your Own Key

The best use of OpenRouter's free tier — extend Claude Code with real-time research for /bin/bash.

✓ With Your OpenRouter Key

  • • Full speed, zero LLM cost
  • • Use any free-tier model
  • • 1 credit per query
  • • Your key, your control

No Key? No Problem

  • • Still works out of the box
  • • 2 credits per query
  • • We handle the LLM
  • • Upgrade anytime

Unlike Tavily or Perplexity API, ContextWire doesn't lock you into their LLM. Plug in your own provider and pay nothing for inference.

Minimal Context, Maximum Signal

Other APIs dump entire pages into your context window. ContextWire gives you exactly what you need.

MetricContextWireTavilyGoogle CSEPerplexity
Avg context per result~150 tokens~800 tokens~200 tokens~2,000 tokens
10-result search payload~1.5K tokens~8K tokens~2K tokens~20K tokens
Context controlFull (profiles, engines, fields)LimitedNoneNone
Custom engine selection105 engines, 22 profiles1 engine1 engine1 engine
Fact verificationBuilt-in (94.3% accuracy)NoNoNo

Build your own custom searches across 100+ engines. Choose exactly which sources matter for your use case — academic, code, news, social, finance, and more.

Simple pricing

Start free. Scale as you grow. BYOK = bring your own OpenRouter key for maximum value.

Free

$0/mo

  • 1,000 queries/month
  • BYOK: 1 credit/query
  • No key: 2 credits/query
  • All 22 search profiles
  • Page extraction
  • MCP server access Get Free API Key

Pro

$10/mo

  • 3,500 queries/month
  • Everything in Free
  • Batch search (parallel)
  • Deep research mode
  • Priority routing
  • Email support Coming Soon

Business

$30/mo

  • 15,000 queries/month
  • Everything in Pro
  • Custom search profiles
  • Dedicated support
  • Usage analytics
  • SLA guarantee Coming Soon

Get your free API key

Start building with ContextWire in seconds. No credit card required.

Create Free Account

相關伺服器