Infrabase
Arvid (independent, Sweden) · Coding
Infrabase is a hand-written directory of 312 AI infrastructure products, covering inference APIs, vector databases, agent frameworks, and observability tools, with a free JSON API.
About Infrabase
What Is Infrabase
Infrabase is a directory of AI infrastructure tools, the parts you assemble around a model rather than the models themselves. That means inference APIs, vector databases, agent frameworks, and observability platforms, the layer where teams spend real money and time. As of late 2026 it lists 312 products, and each entry was written by hand.
That hand-written detail is the point. Most AI directories scrape vendor homepages and republish whatever the marketing copy said the day the crawler ran. Infrabase checks pricing, headquarters country, GDPR status, and license against the vendor's own site, and anything that can't be confirmed gets left out instead of guessed. The trade-off is a smaller, slower-moving list than an automated competitor.
The directory is run by Arvid, a developer in Sweden, working alone. There's no "best" field and no editorial ranking. The comparison pages place sourced facts side by side and stop there, since which tool wins depends on constraints only the reader knows.
Getting Started
- Open infrabase.ai and browse by category, or use the search box for a product, a job like "llm-gateway", or a country.
- Filter the list by category, primary job, GDPR status, or headquarters country from the on-page controls.
- Read a product page for pricing, HQ country, license, and what the tool actually does.
- Query the same data as JSON by calling the /api/query endpoint with parameters like q, category, job, limit, and offset.
- For agent workflows, connect the MCP server at infrabase.ai/mcp from a client such as Claude Code or Cursor, with no key required.
Product Information
A quick look at Infrabase's pricing, supported platforms, and performance.
Best for
The users, tasks, and scenarios where this tool fits best.
Users
- Backend and platform developers
- AI engineers picking a stack
- EU-focused teams
Tasks
- Comparing inference APIs
- Feeding an AI agent
- Checking a vendor's license
Scenarios
- Early research on a new category
- Building an internal tool picker
- Vetting a tool for compliance
Key features
Hand-Verified Listings
A person reads every product page before it appears, checking pricing, license, HQ country, and GDPR status against the vendor's own site. Arvid has caught errors this way, like a provider whose cheapest listed price turned out to be its speech-to-text model rather than an LLM. You get fewer stale entries, at the cost of a shorter list.
A Free JSON API
The whole directory is queryable at /api/query with no key and no signup. You can filter by category, job, GDPR status, or HQ country, or search across names and descriptions, and page through results with limit and offset. The data comes back with per-product fields like tagline, pricing, hq_country, and github_stars.
An MCP Server for Agents
Infrabase runs a remote MCP server at infrabase.ai/mcp that clients such as Claude Code and Cursor can connect to with no authentication. Six read-only tools handle search, single lookups, alternatives, model prices, and a European provider list. It's the same data as the site, shaped for an agent to call mid-task.
Alternatives and Comparison Pages
Every product has an alternatives page, and Infrabase builds head-to-head comparisons that put sourced details next to each other. The pages don't pick a winner by design, since the site has no "best" field and no editorial order. They exist to shorten the gather-your-facts step, not to make the call for you.
Verified Model Prices
A separate dataset tracks per-token model prices across hosting providers, listed cheapest first, and answers questions like the cheapest EU host for a given model. Prices are checked against providers rather than scraped from an aggregator, which keeps them usable for cost planning.
Free, Open Listings
Submitting a product is free and stays free, and featured placement is the only paid option. Featured listings float a product to the top of its category and its alternatives pages for $29 a month or $290 a year. Affiliate links exist on some outbound links but, per the site, don't affect whether or how a product is listed.
Pros and cons
Pros
- Every listing is human-checked against the vendor's site, so license and pricing fields hold up better than scraped directories.
- The JSON API and MCP server are free, key-free, and usable inside agent workflows today.
- EU-focused data like HQ country and GDPR status is verified per product.
- No paid ranking: featured listings sit at the top but don't reorder the rest.
Cons
- The list is smaller than bigger aggregators, so niche or brand-new tools may not be covered yet.
- Data still goes stale between checks, and only some pricing is kept current.
- There's no rating or "best" score, so you won't get a recommendation, just facts to weigh yourself.
- Coverage skews toward infrastructure and EU providers, which is a narrow slice of the wider AI tool market.
Frequently asked questions
Infrabase is a directory of AI infrastructure products, such as inference APIs, vector databases, agent frameworks, and observability tools. Each of the 312 listings is written and verified by hand.
Related content
Explore related tools, skills, and articles for Infrabase.
Infrabase Alternatives
Base Code
Base44 (Wix) · Coding · ProductivityBase Code is a shared cloud development environment from Base44. Teammates describe changes to an existing GitHub codebase in chat, preview them live and send them back as pull requests.
Google EmbeddingGemma 2
Google DeepMind · Writing · Image · CodingGoogle EmbeddingGemma 2 is a 740M-parameter open model that turns text, code, images, audio, and video frames into vectors in one 768-dimension space, running on phones and laptops.
GitGlow
GitGlow · Coding · ProductivityGitGlow is a local-first desktop Git client that lets independent developers review their own code and their AI agents' changes in one place before shipping.