EU-built AI platform with an AI gateway
INFRO vs Orq.ai
Orq.ai is an Amsterdam-built platform for enterprise AI: an AI gateway and an MCP gateway beside AI Studio's prompt management, evaluators, experiments, knowledge bases and managed agents, with an agent-driven Software Factory in early access. Its gateway reaches 30+ providers on Orq's own credits or on your provider keys, with routing rules, fallbacks, load balancing, an exact-match cache, guardrails and budgets at six levels. It is hosted in the EU by default and deploys in your own AWS or Azure account, on-premises from a single Helm chart, or air-gapped. For a European enterprise that wants one platform to build, evaluate and govern AI applications inside its own perimeter, it is a strong fit.
- You want to build AI applications on the same platform that governs them: managed prompts, evaluators, datasets and experiments, knowledge bases and agents. Orq's AI Studio covers that lifecycle. INFRO builds none of it, and sits under whatever you use for it.
- Processing has to stay in the EU, or the platform has to run in your own cloud account, on-premises or air-gapped. Orq is EU-hosted by default and documents all three deployments. INFRO is a hosted service; its gateway runs in Frankfurt beside its database, which is not a residency control, and self-hosted or hybrid deployment is scoped engineering under an enterprise agreement — nothing is packaged for it today.
- You want the platform to fund inference for some workloads and your own keys to serve others, on one account. Orq offers both. INFRO only ever routes to provider accounts you hold.
- Guardrails weigh more in the decision than spend control: secret detection, a model judging requests or responses against criteria you write, custom Python checks, and external policy engines. INFRO's guardrails are patterns and prompt classifiers on chat, on the Business plan.
This section is not a courtesy. If one of these describes you, pick Orq.ai — a customer who adopts the wrong layer on our advice is worse for us than one we never had.
Where the two actually diverge.
- A platform for building AI, or a control plane for operating it
- Orq's gateway is the entry point to a larger product whose job is building AI applications: prompts, evaluation, retrieval, agents and, now, agents that write software. INFRO builds nothing on top of your models. It sits between your applications and the provider accounts you already hold, and its whole surface is operating those calls: which project may call which model, what it may spend, which provider serves it when another fails, and what the request cost — across text, images, video and audio. If you are choosing one vendor for the whole AI lifecycle, theirs is the broader platform. If your teams already build with tools they chose, a control plane under all of them is the narrower problem, and the one INFRO solves.
- What the platform charges for
- Orq has two arrangements. On Orq's credits, a platform fee applies to top-ups. On your own keys, billing stays with your provider and routing is free up to a monthly request allowance, after which a percentage fee applies — read their current terms for the numbers. Either way, what you pay Orq for routing rises with your usage. INFRO has only the second arrangement and takes 0% of model spend: your providers bill you at your own contract rates, and INFRO bills a published subscription for the control plane that does not move with your model bill.
- Estimated cost, and then the provider's own number
- Orq computes each span's cost from per-token rates stored per model in its catalog, with long-prompt tiers applied automatically. Its cost-tracking documentation describes no way to enter your own contract rates and no check against what the provider actually billed. INFRO estimates every request at list rates, or from the Team plan at the rates your contracts carry, per model and provider; on the Business plan it reads OpenAI's and Anthropic's own cost APIs, stores the reported cost beside its estimate for the same days, and states the variance. Where no price is known, the request is marked unpriced, never counted as $0.
- What the budget does at the limit
- Orq's budgets sit at six levels — workspace, project, identity, API key, provider and model — and every matching budget is enforced, returning 429 with a code naming the limit that was hit. INFRO's budgets sit at organization, project, member and key, and refuse with 402 before the provider is called; from the Team plan one cap covers every end customer of your product, with a refusal your application can show its own user. Orq's provider and model levels have no budget equivalent in INFRO, where model and provider access is set by allowlists instead.
Side by side.
Our column carries dashes too. A comparison where one side is all checks is one nobody believes, and rightly.
| Capability | Orq.ai | INFRO |
|---|---|---|
| One API across many providers | yes30+ providers | yesthe providers you connect |
| Your own provider accounts | yesBYOK, or Orq's credits | yesevery request; INFRO holds none |
| Routing, fallbacks and load balancing | yes | yes |
| Budgets that refuse a request | yes429; six levels incl. provider and model | yes402 before the call; org, project, member, key |
| Caps per end customer | yesidentity budgets | yesTeam plan |
| Estimates at your own rates | nocatalog rates; custom rates not documented | yesstored per organization; Team plan |
| Reconciliation against the provider's own bill | no | partialOpenAI and Anthropic, Business plan; variance stated |
| Guardrails | yesPII, secrets, LLM, Python, external | partialchat patterns plus prompt classifiers; Business plan |
| MCP gateway | yesaggregates servers; OAuth sign-in | partialproxies your own servers; Business plan |
| Response caching | partialexact match | partialexact match, chat completions |
| Coding agents through the gateway | yesAnthropic Messages API | yesMessages and Responses APIs |
| Prompts, evaluators and experiments | yesAI Studio | nonot built |
| Knowledge bases and managed agents | yesAI Studio | noroutes your agents' model calls |
| Image, video, and audio | partialimages and audio; no video documented | yesone request shape each |
| EU-hosted by default | yesEU data centers | nogateway in Frankfurt; not a residency control |
| VPC, on-premises and air-gapped deployment | yesAWS, Azure, Helm, air-gapped | partialenterprise agreement; not packaged today |
| Available today | yes | yesself-serve sign-up |
Our reading of each product's public documentation as of October 2026. We do not quote competitors' prices — theirs change and a wrong number about someone else's business is worse than none. Think a row is wrong? Tell us and we will fix it.
Work out what the difference is worth
Estimate what your current model mix costs at your providers' list rates and the plan it needs, or have us read your real usage and tell you where we would and would not help.
Already on Orq.ai? Read the migration guide from Orq.ai: the code that changes, what carries over, and what does not.