Where to Check Jev Status and Uptime

Start with TypeSafe AI's own channels: docs.typesafe.ai and whatever status or announcement links they publish there are the authoritative source for Jev, TypeSafe AI's decision model. We don't know of an official uptime SLA to quote, and we won't invent one. If you reach Jev through the Vercel AI Gateway (the access path described in ecosystem documentation), also check Vercel's status page, since the gateway sits in your request path.

This site doesn't monitor Jev's availability. What it can offer is the checklist for "is it them or me?" and the design that makes the answer matter less.

Is Jev down, or is it you?

Before declaring an outage, rule out the usual suspects:

1. Rate limiting. Bursty batch jobs hit limits and look like outages. Errors that clear when you slow down are limits, not downtime; the rate-limits page covers this.
2. Auth and billing. An expired key or an exhausted balance fails every request at once, which feels exactly like an outage.
3. Your own change. A new question version, schema, or client upgrade deployed today is the likeliest culprit. Roll back and retry.
4. Scope. If one pipeline fails and another works, the problem is probably local.

If all four check out and the official channels confirm an incident, you're in fallback territory.

Design the fallback before the outage

Jev is hosted-only, so there's no local copy to fail over to, as the run-locally page explains. Every pipeline needs a pre-decided policy instead, and the production guide owns the details. The short menu:

• Queue and wait for batch work that tolerates delay.
• Degrade to rules where a deterministic floor exists (keyword filters, allowlists).
• Route to a human for anything consequential.
• Fail over to another model where a framework supports it; the cataloged Pydantic AI integration describes typed routing and fallback workflows.

One rule holds in every branch: a pipeline running degraded should never take irreversible actions on its own. Pause the destructive path, keep the reversible one moving.

Frequently asked questions

Is there an official Jev status page?

Check docs.typesafe.ai for TypeSafe AI's current status and incident channels; we don't link a third-party mirror because official sources outrank us.

Does Jev have an uptime SLA?

We haven't seen a published SLA at the time of writing. Ask TypeSafe AI directly if your use case requires contractual guarantees.

How can I tell rate limiting from an outage?

Rate-limit errors correlate with your request volume and ease when you back off; outages fail regardless of volume. The rate-limits page covers backoff.

What should my pipeline do when Jev is unavailable?

Follow its pre-decided fallback: queue, rules floor, human lane, or a second model, and pause irreversible actions until verdicts return, per the production guide.