Tool returned an error, what now?
Read the error code, retry once, check your inputs. Most errors are transient — second attempt usually succeeds.
Last updated: 2026-05-04
Error code cheat-sheet
| Code | Meaning | What to do |
|---|---|---|
| `MODEL_TIMEOUT` | Model took >60s, request killed | Retry once. If repeats, switch model in advanced settings. |
| `RATE_LIMITED` | You're rate-limited | See [Why am I rate limited?](/help/troubleshooting/why-am-i-rate-limited/) |
| `INVALID_INPUT` | Required field missing or malformed | Check the highlighted form fields. |
| `OUTPUT_TRUNCATED` | Hit token cap mid-output | Reduce target length, or increase cap in advanced. |
| `MODEL_UNAVAILABLE` | Specific model down right now | We auto-fall back to a default — but if you forced a model, switch. |
| `CONTENT_FILTER` | Output flagged by safety filter | Rephrase your request. Some sensitive topics are restricted. |
| `INTERNAL_ERROR` | Bug on our side | Retry once. If reproducible, [contact support](https://aicentraltools.com/contact/) with the run ID. |
| `INSUFFICIENT_CREDITS` | You're out of API credits or quota | See /account/billing/. Maybe overage cap hit. |
What to include if you contact support
- Run ID (in the error message footer or /account/runs/).
- Tool slug (e.g. "blog-post-generator").
- Browser + OS if it's a UI bug.
- What you tried already (retried, switched model, etc.).
Why second attempts often succeed
- Brief network blips during the request.
- Short-lived model overload (especially during peak hours UTC 14:00–22:00).
- Rare model outputs that fail post-validation (e.g. malformed JSON output).
We retry behind the scenes with exponential backoff, but we cap retries at 2 to keep latency reasonable. The third-time-lucky retry from the user often crosses the threshold.
Status page
If you see widespread errors, check /status/. When a model provider is down, we surface it there with ETA and any active incidents.
Was this helpful?
Still stuck? Contact support
