One endpoint for every model, with the controls your team owns
For the team that owns the integration between the company's products and the model providers. Runix Router puts one OpenAI-compatible endpoint in front of every provider, with key custody, per-key limits and failover beneath it. Where training or inference data lives in object storage, Runix FS gives the GPUs a file system over the bucket.
One of six solution pages. Statuses are literal: the products named below are in early access or in development, as marked.
The stack for platform teams
- Runix Router
- One compliant endpoint for every model · early access
- Runix FS
- AI-native file system on your object storage · early access
02 · Situation
The situation
Three things this team is usually dealing with when it calls.
Every team holds its own provider keys
Provider keys end up in repositories, CI variables and notebooks, one per team. Nobody can say which key spent what, and revoking one means a redeploy. The platform team is asked to fix this without breaking the teams that depend on it.
A provider outage is your outage
When a provider degrades, every product that calls it directly degrades with it. Failover written into each application is written differently each time, and a failure mid-stream is the case nobody tested. The platform team ends up on call for other people's integrations.
The GPUs wait on the bucket
Training and inference data sits in object storage because that is where it is cheap. Jobs then read it object by object, each framework with its own client for the same bucket, and the GPUs wait while the data arrives.
03 · Stack
The products, and the role each plays
Each one works on its own; together they are one path through the stack.
04 · Start
How it starts
A person on the other end at every step; nothing here is self-serve.
01Tell us what you are building
Send the models you call, the expected volume and what you need from the contract. Router is invite-only during early access; a person replies within one business day and sets the account up after a short intake.
02Evaluate before you pay
Evaluation credits for Router are issued on request, with full functionality and no card required. Each key carries its own quota, limits and allowed models, and billing is itemised, so what each team spends is visible from the first request.
03Scope FS per deployment
If the GPUs read from object storage, Runix FS is sized and deployed with you in your own AWS, Google Cloud or Azure account, and quoted per deployment during early access.
05 · In writing
What is in writing
The parts a review asks for, stated once and linked to the page that holds them.
Data terms
Content sent through Runix Router is processed to serve the request and never used for training. Operational metadata is kept for billing and support.
Contracting and billing
You contract with Runix AI Inc; an MSA and a DPA are available on request. Billing is usage-based in USD, itemised, on a prepaid balance or by invoice.
Statuses, literally
Runix Router is in early access and running traffic, with no published SLA; the failover mechanism is documented instead. Runix FS is in early access, per deployment, and supported under contract.
06 · Reading
Reading for this team
Engineering notes that cover the mechanics behind this page.
- Migrating to a gateway without a maintenance window
- Streaming LLM Responses Without Breaking Failover
- One key can drain a shared quota pool
Other teams:
07 · Questions
Common questions
Do we have to change our client code?
Runix Router is OpenAI-compatible: chat completions and model listing, with server-sent events for streaming, behind an OpenAI-style error envelope with a request id on every response. You choose a model with an explicit id, or with auto.
What happens when a provider fails mid-stream?
Failover is provider to provider and fires mid-request, with streaming preserved. The mechanism is documented under Reliability; there is no published SLA during early access.
Does Runix FS replace our object storage?
No. In cache mode the bucket is the source of truth and writes pass through; in file-system mode operations complete in the cluster and then sync to the bucket. Paths map one-to-one to object keys, so the bucket stays independently readable.
Put one endpoint in front of every provider
Tell us what you run today and what you are trying to change. We reply within one business day with a concrete plan, and we say which parts we would not do.
Talk to sales