Live prices and specs of 400+ AI models from every major provider (OpenAI, Anthropic, Google, Meta, Mistral, DeepSeek, Qwen...): USD per million input/output tokens, cached-input price, context length, max output, and features (tools, JSON, reasoning, vision). Filter by ?q=, provider, needs=tools,vision, minContext, maxInputPrice; sort by price, context or newest.

Cheapest AI model that can do your job. Describe the task (or list needed features: tools, json, reasoning, vision...), expected input/output tokens and a quality level (budget/balanced/best); get the cheapest suitable models across all major providers with estimated USD cost per call and per 1,000 calls, from live prices. POST {task, inputTokens, outputTokens}.

Blockchains ranked by total value locked in DeFi or by US-dollar stablecoins circulating on them, with both figures and each chain's gas-fee currency symbol (e.g. ETH, SOL). Up to 50 chains. DefiLlama data; for one chain in depth use chain-stats.

One blockchain's economy in one call: total value locked, DEX trading volume (24 h, 7 d, 30 d and the day-on-day change), its five busiest DEXs, fees paid on the chain and the stablecoins circulating on it. Ethereum, Solana, Base, Arbitrum, BSC and 400+ more. Chain-level figures from DefiLlama; nothing about individual wallets.

What changed in a library or SDK since a date: every version published since then (npm, PyPI, crates, Go) and the release notes from its GitHub releases or CHANGELOG, newest first, with breaking-looking lines flagged. Works for any GitHub repository too (API SDKs, CLIs). ?name=openai&since=2026-09-01 or ?repo=stripe/stripe-node&since=2026-09-01

Pay-per-call LLM chat (Mistral Small 3.1 24B): Balanced multilingual model (140+ languages) for writing, reasoning and long inputs. No API key or account. POST JSON {"prompt": "..."} or OpenAI-style {"messages": [{"role": "user", "content": "..."}]}, optional "system", "max_tokens" (up to 1024), "temperature", "json": true. Up to about 20,700 characters of English in. Failed calls are not charged.

Pay-per-call LLM chat (Llama 3.2 3B): Fast, cheap model for classification, extraction, short answers and routing. No API key or account. POST JSON {"prompt": "..."} or OpenAI-style {"messages": [{"role": "user", "content": "..."}]}, optional "system", "max_tokens" (up to 1024), "temperature", "json": true. Up to about 15,000 characters of English in. Failed calls are not charged.

Pay-per-call LLM chat (Llama 3.3 70B): Strongest model for harder reasoning, coding and careful writing; supports JSON output. No API key or account. POST JSON {"prompt": "..."} or OpenAI-style {"messages": [{"role": "user", "content": "..."}]}, optional "system", "max_tokens" (up to 1536), "temperature", "json": true. Up to about 36,000 characters of English in. Failed calls are not charged.

Live company profile from its domain: everything in /company/techstack plus domain age and registrar, security headers, likely legal entities in the global LEI register, and UK Companies House matches (when enabled). Companies only, no personal data. Pass ?domain=stripe.com&name=Stripe. For an AI-written report with news use /company/report.

Company tech stack lookup (what technology does a company use?) from its website and DNS: CMS/website builder, frameworks, analytics and ad pixels, marketing and support tools, payments, hosting/CDN, email provider and senders, DNS host and SaaS verification records (Slack, Atlassian, Stripe...), plus official social profiles. domain is the company website (stripe.com); name, if you know it, improves matching. ?domain=stripe.com

Do these package versions work together and on your runtime? Give up to 20 npm or PyPI packages (exact versions, ranges or latest) plus your Node or Python version; each package's engines, peer dependencies or Requires-Python / Requires-Dist are checked against the others and the runtime, with conflicts explained and end-of-life runtimes flagged.

Convert between crypto and fiat at live rates: BTC, ETH, SOL, USDC and hundreds more against USD, GBP, EUR and 150+ currencies, either way, or crypto to crypto. Pass ?amount=0.5&from=BTC&to=GBP. Rate from Coinbase, cross-checked against Kraken's last trade where Kraken lists the pair, with the gap shown.

Turn a small website, or one section of it, into clean Markdown: starts from your URL, finds pages from the sitemap or by following links on the same site, obeys robots.txt, and returns up to 10 pages (title, Markdown, address) in one call. Great for docs, blogs and company sites. ?url=https://example.com/docs/&maxPages=10&path=/docs/

Validate and explain a cron expression: plain-English meaning, each field expanded, and the next run times in any time zone (DST-aware). Standard 5-field cron, 6-field with seconds, names (MON, JAN) and @daily-style macros. ?expression=*/15 9-17 * * MON-FRI&tz=Europe/London

Vulnerability lookup for agents: by id (?id=CVE-2021-44228, GHSA-..., PYSEC-..., RUSTSEC-..., GO-...) get the advisory, severity/CVSS, affected packages and fixed versions, exploit probability (EPSS) and whether CISA lists it as exploited in the wild; or by package (?ecosystem=npm&package=lodash[&version=]) list every advisory with the same enrichment.

What DAOs are voting on right now: open Snapshot proposals from verified DAOs, biggest communities first, each with the DAO, title, closing time, voters so far, current vote shares and leader, quorum and link. A daily governance watch for agents and funds. Filter by minimum followers. No voter addresses.

A DAO's governance votes from Snapshot: the latest proposals (active, closed or all) with title, dates, voting type, each choice's votes and share, the leading choice, number of voters, quorum and whether it was met, a short summary and the link. Give the Snapshot space id, e.g. uniswapgovernance.eth, aavedao.eth, arbitrumfoundation.eth. No voter addresses.

Find a DAO's Snapshot space by name: returns matching spaces, biggest first, with their id (to use with dao-proposals), whether Snapshot has verified them, followers, number of proposals and how many votes are open now. E.g. q=uniswap, aave, arbitrum, ens.

Recent crypto hacks and exploits (DeFi, bridges, exchanges) from DefiLlama's database: date, amount, technique, chains, funds returned, with totals by type, plus live alerts for protocols whose TVL is suddenly crashing (possible ongoing exploit). Filters: ?days=30&chain=Ethereum&protocol=...&minAmountUsd=1000000.

DeFi protocol safety check: TVL and its 1h/1d/7d change, chains, category, audits and audit links, oracles, age, forks, market cap/TVL, and every past hack from DefiLlama's database, rolled into flags, a 0-100 score and a verdict. Pass ?name=aave-v3 (slug or name).

Best DeFi yields for a token and chain, ranked by risk-adjusted APY (sustainable APY, discounted for risk): APY (base vs rewards), TVL, 30-day average, impermanent-loss risk, audits and past hacks of the protocol, with a 0-100 risk score and plain risk notes per pool. E.g. ?token=USDC&chain=base&minTvlUsd=1000000. Filters: project, stablecoinOnly, noImpermanentLoss, sort.

"Should I use this dependency?" in one call for npm, PyPI, crates or Go: package safety (vulnerabilities, malware, typosquats, deprecation, licence, downloads) plus its GitHub repository health (activity, bus factor, releases, Scorecard), a decision (use / use-with-care / avoid) with reasons and plain-English advice. name is the package and ecosystem its registry; version is optional (default: the latest). ?ecosystem=npm&name=express

The busiest decentralised exchanges: DEXs ranked by trading volume over 24 hours, 7 days or 30 days, across all chains or on one chain (e.g. Base, Solana, Arbitrum), with each one's share of the total, day-on-day change and number of chains. Up to 50 rows. DefiLlama data.

Look up DNS records for any domain: A, AAAA, CNAME, MX, NS, TXT, SOA, CAA, SRV, PTR, DS, DNSKEY, HTTPS, or ALL common types at once. Pass ?name=example.com&type=MX. An IP address with type=PTR does a reverse lookup. Uses DNS-over-HTTPS (Cloudflare 1.1.1.1) and reports DNSSEC validation and NXDOMAIN.

Should you build on this Docker Hub image? Checks official status, deprecation, when the tag was last rebuilt (stale images miss security patches), mutable tags like latest, pulls, architectures and size, and whether the runtime or OS version in the tag is end-of-life (endoflife.date). Verdict and reasons. Not a CVE scan. ?image=node:18-alpine

Reads the project's current docs (llms.txt, docs pages or latest README) for any npm, PyPI, crates or Go library and answers from them alone, with a code example, the version it applies to and cited sources; says so when the docs don't cover it. library is the package name, question your question; ecosystem defaults to npm. POST {library, question}.

Find the AI-ready docs for any library or API: checks the project's docs site (from its npm, PyPI, crates or Go metadata) or any company domain (docs., developers., /docs) for llms.txt and llms-full.txt files, with titles, sizes and page counts, plus homepage and repository. ?library=hono or ?site=stripe.com

Up-to-date docs for any npm, PyPI, crates or Go library: finds the project's own llms.txt and docs pages (or the latest release's README), keeps only the sections relevant to topic within a tokens budget, and returns them with sources and the current version. library is the package name; ecosystem defaults to npm. ?library=hono&topic=middleware&tokens=3000

Check an email address before you use it: syntax, whether its domain has mail servers (MX, with null-MX and A-record fallback), disposable/throwaway providers, role accounts (info@, support@), free providers, and typo suggestions (gmial.com → gmail.com). Pass ?email=someone@example.com. Does not contact the mailbox.

Text embeddings for semantic search, clustering and RAG: 1024-dimension multilingual vectors (BGE-M3, 100+ languages). POST JSON {"input": ["first text", "second text"]} (up to 32 texts of 8,000 characters) or {"text": "one text"}. One price per call, however many texts. No API key needed.

Explain and fix an error: finds matching GitHub issues and Stack Overflow questions (like /error/known), reads the top accepted answers in full, then writes the likely cause and concrete fix steps citing those sources, and says plainly when nothing matching was found. Paste the error (stack traces fine) and optionally the library.

Is this error known? Paste an error message or stack trace (and optionally the library): it is cleaned of paths and line numbers, then searched in the library's own GitHub issues, all GitHub issues and Stack Overflow, returning the best matches with status (closed, accepted answer...), links and excerpts. For a likely fix use /error/explain.

Official ECB euro foreign-exchange reference rates for about 30 currencies, converted to any base (GBP, USD, EUR...). Latest or any past date since 1999 (the last rate published on or before it). Optional amount conversion.

Look up legal entities in the global LEI register (GLEIF, 2.9M+ companies and funds worldwide): by LEI (with direct and ultimate parent companies), by name, or by national company number. Registered address, status, jurisdiction. Companies only.

Check an EU VAT number live against the European Commission's VIES system: valid or not, and the registered company name and address where the member state shares them. Northern Ireland (XI) numbers included. Sole traders' names are withheld (personal data).

Web page extractor for AI agents: fetch any public URL and get clean Markdown, the page title and all links as JSON, with scripts, styles and menus removed. url is the page address (https://...). Private or internal addresses are blocked; pages over 2 MB or slower than 10 seconds are rejected and you are not charged.

AI structured extraction: give a URL (web page, PDF, Word, Excel...) or text plus your own JSON Schema, get back JSON that fits it. POST {"url": "...", "schema": {"type": "object", "properties": {...}}, "instructions": "optional"}. Great for prices, contacts, specs, events, invoices and tables. Uses Llama 3.3 70B in JSON mode. Up to 40,000 characters of source; failures are not charged.

Live network fees on Ethereum, Base, Arbitrum, Optimism, Polygon, BNB, Avalanche, Solana and Bitcoin: slow/standard/fast fee rates, congestion and trend, and the USD cost of a send, token transfer and swap on each, plus the cheapest chain now. Pick chains with chains (e.g. base,solana); all by default.

Is this GitHub Action safe for your workflow? Pass what follows uses: (e.g. tj-actions/changed-files@v45). Checks known advisories and compromises (OSV), SHA vs tag vs branch pinning, deprecated Node runtimes, unpinned Docker images and nested actions in action.yml, publisher and repository health. Verdict, score and fixes.

Describe an image for an AI agent: a caption (short) or a detailed description of the scene, objects, text, colours and layout, plus alt text. JPEG, PNG, GIF or WebP up to 5 MB from a public URL. People are described generically, never identified.

Read the text in an image (OCR): screenshots, photos of signs, receipts, slides, scanned pages. Returns the text in reading order with line breaks, and whether any text was found. JPEG, PNG, GIF or WebP up to 5 MB from a public URL.

Licence compatibility check for coding agents: give your project's licence (or proprietary) and how it's used (distributed, SaaS or internal) plus up to 50 dependencies as SPDX ids or packages (npm:express@4.21.2); get ok / conditions / review / incompatible per dependency with the reason (GPL, AGPL, LGPL, MPL, Apache, dual licences...). Not legal advice.

Convert any web page, PDF, Word (.docx), Excel (.xlsx/.xls), OpenDocument, CSV or XML URL to clean, LLM-ready Markdown, up to 10 MB: returns title, Markdown, links (web pages) and character count. url is the document address (https://...). For JavaScript-heavy sites use /markdown/rendered. Private addresses are blocked; failures are not charged. ?url=https://example.com/report.pdf

Scrape JavaScript-heavy web pages (single-page apps, React/Vue sites, dashboards) to clean Markdown in a real headless browser: scripts run, then the rendered page is converted. Returns title, Markdown and links. url is the page address (https://...). Use /markdown for ordinary pages and PDFs (cheaper). Private addresses are blocked; failures are not charged.

Prediction-market odds from Polymarket: search any topic (?q=bitcoin, fed rates, election, a team) or omit q for the most-traded markets now. Each event lists its markets with outcome probabilities (%), 24h change, volume, liquidity, bid/ask, end date and link. For research, forecasting and trading agents.

A web page's metadata in one cheap call: title, description, canonical URL, language, site name, preview image, OpenGraph and Twitter card tags, author and dates, icons, RSS/Atom feeds and JSON-LD (schema.org) types. For link previews, SEO checks and crawlers. robots.txt respected. ?url=https://example.com

Research any topic across Hacker News and headlines from major outlets (BBC, Guardian, Ars Technica, TechCrunch, crypto news and more), Lobsters, Lemmy and DEV in one call: recent stories merged, de-duplicated and sorted newest first, with links, discussion threads, points, comment counts and outlets. q is the topic; days (1-30, default 7) is how far back to look; sources picks which feeds to search. For an AI-written summary with citations use /news/brief. ?q=topic&days=7

AI research brief on any topic from the last 1-30 days: gathers Hacker News and headlines from major outlets, Lobsters, Lemmy and DEV, then writes a concise briefing (key developments, sentiment, what to watch) with numbered citations to the sources, plus the full item list. Pass ?q=topic&days=7. Built for research agents.

Understand any public API fast: give its OpenAPI/Swagger spec URL, its docs or base URL, or a name from the APIs.guru directory (?api=stripe.com); we find the spec and return title, version, servers, auth schemes, tags and every operation (method, path, summary, parameters, body, responses), filterable by tag or search. ?api=stripe.com&q=customer

Audit a whole dependency list in one call: up to 200 exact npm, PyPI, crates or Go package versions checked against OSV.dev for known vulnerabilities and malware. Returns which packages are affected, severity, fixed versions and an overall verdict. POST {ecosystem, packages:["name@version"]} or {ecosystem:"pypi", requirements:"requests==2.31.0\n..."}.