Get a key when the gateway opens.
INFRO is in build. Early accounts get keys first, and get asked what to prioritise — the catalog and the full API contract are already public so you can review both before you commit anything.
- Read the contract first. Every endpoint is documented — request shape, error taxonomy, webhooks — before it ships, so you can tell us it is wrong while that is still cheap.
- No trial credits. The wallet is strictly prepaid, which is what keeps the platform free of the fraud that funds a free tier. You will pay for what you run.
- Two lines to try it. Text keeps the OpenAI request shape, so evaluating it means pointing an existing client at a new base URL — not a rewrite.
What will exist when you get that key
The whole ledger, including what we have not built. Nothing on this site claims a capability that is not on this list at the status shown here.
Available now
Working today. Open a tab and check.
- Model catalog
- 150+ text, image, video, and audio models with reference and INFRO pricing, browsable now at /models.
- API documentation
- The full request and response contract for every endpoint, published before the endpoints exist so it can be reviewed first.
At launch
In the first release. Built or being built now — you cannot use it yet.
- Text — OpenAI-compatible
- POST /v1/chat/completions with the request shape the OpenAI SDKs already send, including streaming and tool calling.
- Image, video, and audio
- One request shape per modality across every model, with video and long audio running as async jobs.
- Async jobs and webhooks
- Submit long renders, get a job id immediately, receive a signed webhook on completion.
- First-party SDKs
- Python, TypeScript, and Dart clients covering every modality on one key. The OpenAI SDKs work against the text endpoint without them.
- Media re-hosting
- Generated media is copied to INFRO storage so output URLs do not expire underneath your application.
- Health-aware routing
- Requests are scored against live error rate, latency, and rate-limit pressure before a route is chosen.
- Automatic failover
- A route that times out or 5xxs before the response starts is retried elsewhere without your client seeing it.
- Usage and cost per request
- Every request records its model, latency, token or unit counts, and exact cost, queryable by project and key.
- Request traces and payload logs
- Per-request timing detail, with prompt and completion logging off by default and a retention window you choose.
- CSV and usage API
- Request-level export so the numbers land in your warehouse.
- Budgets and spend ceilings
- A hard cap per organization, project, member, and key, with alerts at your own thresholds.
- Spend attribution
- Group cost by model, project, key, or an end-user id you attach to the request.
- Organizations and roles
- Owner, admin, developer, and analyst roles with keys and console access scoped to each.
- Model allowlists
- Approve which models a project may call; anything else returns a policy error instead of a line item.
- Audit log
- Member, key, and policy changes recorded with actor and timestamp, exportable as CSV.
Enterprise
Real, but sold with an agreement rather than self-serve.
- SSO and SCIM
- SAML and OIDC sign-in with SCIM provisioning, so access follows your identity provider.
- Contractual SLA
- An availability commitment with remedies, agreed per contract. There is no published SLA before the gateway has an operating history.
On the roadmap
Deliberately after the first release. No date, no commitment.
- Bring your own key
- Route a provider's traffic through your own contract and rates while keeping INFRO's routing and analytics.
- OpenTelemetry export
- Traces pushed to the stack you already run rather than read in our console.
- Console playground
- Run any catalog model from the browser and copy the equivalent API call.
- Cost recommendations
- Automatic analysis of where a workload could move to a cheaper equivalent model.
- Spend change analysis
- Attributing a month-over-month bill increase to request volume, output length, model mix, and new workloads.
- Regional processing
- Pinning an organization or project to a region requires infrastructure we have not bought yet. We will not claim it before it is enforced.
Legend:Available now — Working today — you can check this yourself.At launch — Committed to the first release. In build now, not usable yet.Enterprise — Available on an enterprise agreement rather than self-serve.On the roadmap — Planned after the first release. No date committed.
Spending real money on inference already? Ask for a cost audit instead — we will price your actual workload here before you write any code.