Many people still think of You.com as a consumer AI search engine, but its own site now presents it as a provider of search, answer and research APIs for AI agents and enterprises. You.com is a company that sells web search, page contents, answer and research APIs, plus an Inference Router for open-weight models, to developers and enterprises; its Series C announcement says it began connecting large language models to live web data with citations in December 2022. This page sets out what You.com offers as of November 2026, how its APIs retrieve and cite sources, what each API costs, how access and keys work, which security commitments it states and how the product has changed. It is written for developers, architects, procurement and security teams evaluating a search API, and for anyone asking what happened to you.com.
Last checked: 2026-11-15.
This is an independent guide from NHI Mgmt Group, based on You.com's public documentation as of November 2026. It is not sponsored or reviewed by You.com. Product features change, so confirm settings against You.com's current documentation.
Key takeaways
- An API company now. As of 15 November 2026 the you.com homepage describes the Web Search, Contents, Answer, Research and Finance Research APIs, an Inference Router, an MCP server and agent skills, all aimed at developers and enterprises.
- Pay as you go pricing. The docs list $5.00 per 1,000 Web Search calls, $5.00 per 1,000 Answer calls, $1.00 per 1,000 Contents pages and Research API tiers from $12 to $1,200 per 1,000 calls.
- Citations are built in. The Answer API returns a Markdown answer with citations carrying verbatim excerpts and the web results behind it; the Research API returns cited Markdown or structured JSON.
- Keys, caps and keyless options. Requests use an
X-API-Keyheader; per-key spending caps arrived on 5 October 2026; agents can also pay per request through x402 or the Machine Payments Protocol. - Stated security commitments. The homepage lists SOC 2, zero data retention for search, a DPA and CCPA compliance, and describes the Inference Router as US-hosted with a 99.9% uptime SLA.
What is You.com?
You.com describes its products as infrastructure for agents and applications that need live web data. A developer sends a query or a list of URLs and gets back ranked results, extracted page text, a cited answer or a multi-step research report. Its Series C post (3 September 2025) says that customer requests to connect models to live web data led to its APIs and its shift towards AI infrastructure.
| Field | Value (as of 2026-11-15) |
|---|---|
| Provider | You.com; CEO and founder Richard Socher, per the About page |
| Origin of web-grounded answers | December 2022, per You.com's Series C post |
| Current focus | Search, contents, answer and research APIs, Inference Router, MCP server |
| Consumer search | None described on the you.com homepage |
| Funding (latest stated) | $100 million Series C at a $1.5 billion valuation, led by Cox Enterprises, 3 September 2025 |
| Usage reported by You.com | Over 1 billion queries a month through its API (provider figure, September 2025) |
| Customers named by You.com | DuckDuckGo, Harvey and Windsurf (Series C post) |
| Price | Free API key; pay as you go; Enterprise custom pricing |
| Ads | Not stated by You.com |
How do you access You.com?
- Platform: developers create a free API key at you.com/platform. Enterprise plans with custom pricing, higher rate limits and dedicated support are arranged through a demo request.
- Authentication: the key goes in the
X-API-Keyheader, and the docs use theYDC_API_KEYenvironment variable in examples. A key is shown only once, at creation. - Roles: Owners and Admins see every key in the organisation; Developers see only keys they created.
- MCP: a hosted server at
https://api.you.com/mcpwith bearer token or OAuth 2.1 sign-in, plus a local package,npx @youdotcom-oss/mcp, over STDIO. - Keyless payment: since 4 August 2026, x402 and the Machine Payments Protocol can pay for Web Search and Finance Research requests per call from a funded wallet. The docs say keys are still needed for zero data retention, usage analytics and team roles.
- Rate limits: the docs describe rate limit headers, HTTP 429 responses with
Retry-Afterand exponential backoff; a key that hits its spending cap returns 429 with the codespend_cap_exceeded.
How does You.com find and cite sources?
Each API returns sources in a different shape. The Web Search API is the base layer: POST https://ydc-index.io/v1/search, returning web and news results in one results.web array, with news shown when the query has news intent.
| API | Use case | Sources returned | Latency stated |
|---|---|---|---|
| Answer | Cited answer from one search pass | Markdown answer, citations with verbatim excerpts, web results | p50 2.67 seconds |
| Contents | Page text from URLs you already have | Markdown, HTML, metadata | Up to crawl_timeout per URL, default 10 seconds |
| Finance Research | Filings, prices, fundamentals and financial news | Markdown answer plus sources | Under 120 seconds (deep) or 300 seconds (exhaustive) |
| Research | Multi-step cited answers with depth control | Markdown or structured JSON plus sources | Under 10 seconds (lite) to background only (frontier) |
| Web Search | Ranked results for your own model, RAG pipeline or interface | URL, title, description, snippets, highlights or full page, page age | Retrieval only, faster than Answer |
Web Search request controls stated in the docs:
- Volume:
countdefault 10, maximum 100 results per call;offset0 to 9 for pagination. - Scope:
freshness(day, week, month, year or a date range),country,languageas a BCP 47 code. - Safety:
safesearchoff, moderate (default) or strict. - Domains:
include_domainsup to 500,exclude_domainsup to 5,000,boost_domainsup to 500. - Page content:
extractionwith highlights or full page, sourced from cache, live fetch or a blend (default). - Knowledge: results from licensed data providers, requested with
knowledge: "core", at no extra charge.
The docs do not give an index size or refresh rate.
Which models power You.com?
The search, contents and answer APIs do not name the model behind them in the pages fetched for this guide. The Inference Router, launched on 26 October 2026 according to the changelog, gives access to open-weight models through one OpenAI-compatible endpoint (POST /v1/chat/completions), with an Anthropic-compatible Messages endpoint described in the docs index.
| Item | What You.com states |
|---|---|
| Models listed on the homepage | DeepSeek V4.1 Flash, GLM 5.3, Inception Mercury 3, Kimi K3, Muse Spark 1.3, Qwen3.8 Flash |
| Pricing | Set per model; no platform fee |
| Reliability | At least 99.9% uptime SLA with failover to backup providers |
| Data handling | Zero data retention across models and web search; 100% US-hosted |
| Web grounding | Built-in web tools ground answers in live web results |
As of 2026-11-15. Models are listed alphabetically. For how open-weight and closed models differ, see Open-Weight vs Closed Models Explained.
How much does You.com cost?
All API prices are pay as you go; a free API key gets a developer started and Enterprise pricing is custom. As listed in the docs on 2026-11-15:
| API | Price per 1,000 |
|---|---|
| Web Search | $5.00 per call; full page add-on $1.00 per live page |
| Contents | $1.00 per page |
| Answer | $5.00 per call |
| Research | lite $12, standard $50, deep $100, exhaustive $450, frontier $1,200 per call |
| Finance Research | deep $110, exhaustive $500 per call |
| Inference Router | Varies by model, no platform fee |
How do ads work in You.com?
Not stated by You.com. Its APIs are sold on a pay as you go basis per call or per page, and the docs and homepage fetched for this guide do not describe advertising in API responses.
What can publishers control?
You.com's documentation fetched for this guide does not name a crawler user agent or a publisher opt-out for its index or the Contents API. Publishers who want to limit AI crawling in general should use robots.txt, meta tags and headers as described in AI Crawler and Scraper Controls. On the customer side, the exclude_domains and include_domains filters let a developer decide which sites their own application draws on.
What privacy controls exist?
The you.com homepage lists zero data retention for models and web search, with queries and data able to be purged automatically, SOC 2 certification, readiness for Data Processing Agreements and CCPA compliance, and describes its models as US-hosted. The API key docs say keys are needed for zero data retention, so keyless machine payments do not carry it. Per-key spending caps and roles are covered under access above.
When was You.com released, and how has it changed?
| Date | Milestone |
|---|---|
| December 2022 | Began connecting large language models to live web data with citations, after ChatGPT's launch (per the Series C post) |
| 12 August 2025 | Migration post for developers after Microsoft retired Bing Search APIs on 11 August 2025 |
| 3 September 2025 | $100 million Series C at a $1.5 billion valuation led by Cox Enterprises, with Georgian, Norwest and Salesforce Ventures; over 1 billion API queries a month reported |
| 5 August 2026 | Answer API launched |
| 5 October 2026 | Per-key spending caps |
| 26 October 2026 | Inference Router launched |
As of 2026-11-15. Earlier consumer milestones are not listed because they could not be matched to a You.com page.
What features have been added?
| Date | Change in the You.com changelog |
|---|---|
| 15 July 2026 | authors field removed from Web Search API v1 and Answer API responses |
| 20 July 2026 | Background mode for long Research API tasks and a new frontier effort level |
| 4 August 2026 | Machine payments (x402 and MPP) for Web Search, Finance Research and /v1/agents/search |
| 5 August 2026 | Answer API launched |
| 11 August 2026 | extraction parameter for highlights or full page on /v1/search |
| 2 September 2026 | extraction_source to choose cached, live or blended content |
| 23 September 2026 | Knowledge results from licensed data providers |
| 5 October 2026 | Spending caps for individual API keys |
| 7 October 2026 | Structured JSON output through output_schema for Finance Research |
| 9 October 2026 | total_max_tokens highlight budget on /v1/search |
| 26 October 2026 | Inference Router with OpenAI- and Anthropic-compatible endpoints and web tools |
As of 2026-11-15. The Inference Router launch blog post is dated 22 October 2026.
What are You.com's limitations?
- Costs scale with depth: a Research API call at the frontier tier is priced 100 times a lite call ($1,200 versus $12 per 1,000), and frontier requires background mode.
- No cap by default: new keys have no spending cap until one is set, so an unattended agent can spend up to the organisation's credit.
- Breaking changes happen: the 15 July 2026 removal of the
authorsfield shows that response fields can be withdrawn; pin parsers to documented fields. - Index facts not published: the docs do not state index size, coverage or refresh rate.
- Benchmarks need checking: the homepage links a benchmarks page; read the methodology there before relying on a figure.
Common questions about the service
Is You.com free?
Partly. A free API key is available and the hosted MCP server's ?profile=free exposes you-search and you-discover without a key, limited to 100 queries a day. Beyond that, usage is pay as you go at the published per-call and per-page prices, and Enterprise pricing is custom. The /mcp/research and /mcp/finance endpoints have no free usage.
Is You.com a search engine or an API?
As of November 2026, an API provider. The homepage describes developer and enterprise APIs and no consumer search product. Its results reach end users through other products: the Series C post names DuckDuckGo among its customers, and You.com's resources page reports integrations such as Scira using the Web Search API as its default search provider (29 October 2026).
Where identity and security fit
Every You.com integration runs on a machine credential: an API key, an MCP bearer token, an OAuth 2.1 grant or a funded payment wallet. The docs recommend environment variables over source code, .env files in .gitignore, separate keys per environment, periodic rotation (create, switch, delete; deletion revokes at once) and least privilege. Per-key caps and the MCP tool allowlist (?tools=, X-Allowed-Tools or YDC_ALLOWED_TOOLS; research and finance tools are off by default) limit what a leaked key or an over-eager agent can do. See the API Key Management Guide, API Keys vs OAuth for MCP Servers and the AI model provider API keys guide. For vendor due diligence, the homepage links a trust centre, a DPA and an MSA dated December 2025; the AI Vendor Security Questionnaire Template covers the questions to ask.
Practitioner checklist
- Set a spending cap on every key at creation, starting with keys used by unattended agents.
- Store
YDC_API_KEYin a secrets manager or environment variable, never in code, with separate keys for development, staging and production. - Use an API key rather than keyless machine payments where you need zero data retention, usage analytics or team roles.
- Restrict MCP tools with an allowlist and enable research or finance tools only where needed.
- Choose the cheapest API that meets the need: Contents for known URLs, Web Search for your own synthesis, Answer or Research for cited answers.
- Handle 429 responses by telling
spend_cap_exceededapart from rate limiting, with backoff that honoursRetry-After. - Watch the changelog for removed fields and review the DPA, MSA and trust centre before production use.
Standards and references
- You.com Docs: Choose the right API
- You.com Docs: Web Search API overview
- You.com Docs: API keys
- You.com Docs: MCP server
- You.com Docs: Changelog
- You.com: Homepage, products and trust commitments
- You.com: Series C announcement
- You.com: About
Related NHI Mgmt Group resources: AI Answer Engines Hub · API Keys vs OAuth for MCP Servers · API Key Management Guide