The surface is narrow on purpose: two metered endpoints and model: "auto". Model identity is not exposed.
ModelBeatbeta

Support

How to reach us during private beta, what to include, and what is and is not promised.

ModelBeat is in private beta. There is no published SLA, no uptime commitment, and no support response target. What there is: a small team that reads everything.

Getting help

Reach us by email, during business hours, at hello@neuralarc.ai.

Include the X-Request-Id from the response. It is on every response, success or failure, and it is the fastest path to the exact call in our records — usually the difference between an answer in one reply and four rounds of narrowing down.

X-Request-Id: req_01HXYZ...

See Errors for how to read it off both a success and a raised exception.

What to send

ReportingInclude
A failing requestX-Request-Id, the status code, and the literal error value. Never the API key.
A billing discrepancyThe X-Request-Id and your recorded usage.cost.total_cost for it. See Costs.
A security issueWrite first, before any public disclosure. Do not include live keys in the report.

Never send us a key

No support question needs your mb_live_ key, and we will never ask for one. If a key has been exposed anywhere, revoke it in the console first and reissue. See Authentication.

Requests we prioritise by demand

Several beta constraints are deliberate rather than permanent, and demand is how they get scheduled. Tell us if you need:

  • The served model on the response, programmatically. See Routing.
  • The Anthropic Messages or OpenAI Responses request shape. See Limits.
  • The DPA or the sub-processor list. See Data handling.

What changes without notice

Anything not in the API reference is not part of the API and may change or disappear at any time. The documented surface will change by growing, not by breaking what is written down. See Limits.

On this page