Last updated: July 14, 2026
Trace Support.
If transcription or analysis fails, first check your Gemini API key, project quota, audio format, and foreground processing state. This page summarizes common Gemini API errors, what to try, and the privacy model.
When processing fails
Trace records meetings directly or imports audio from Files and Voice Memos, then uses your own Gemini API key to create transcripts, meeting notes, decisions, risks, and action items.
If processing fails, check the items below first. If the issue keeps happening, you can leave public feedback through an App Store review.
If you do not have a Gemini API key yet, start with the Gemini API Key Guide. If you want to estimate longer-audio processing costs, see Models & Costs.
Before contacting support
Please confirm:
- Trace has microphone permission. If you previously denied access, enable it in iOS Settings.
- You entered a valid Gemini API key in Settings.
- The Google project that owns the key still has available quota.
- The app stays in the foreground for longer audio processing. If iOS interrupts background execution, return to the app and retry the unfinished workflow.
- The imported file is in an audio format supported by the app.
- If a call, another app, or an audio device interrupts recording, return to Trace and confirm the recording state before continuing or finishing.
- If sharing audio or a result fails, reopen that item and use the system Share Sheet again.
Gemini API errors
Trace uses your own Gemini API key and calls the Google Gemini API directly from your device. These errors usually point to Gemini API availability, your Google project, the selected model, quota, billing, or network state. They do not indicate a Trace server outage; Trace does not run a backend for these Gemini requests.
Quick read: For HTTP 500, 503, or 504, first retry after a short wait, switch models, reduce meeting background or input context, and check Gemini API status.
| Error | Common meaning | What to try |
|---|---|---|
| HTTP 400 / INVALID_ARGUMENT | The request format, context, or model parameters were not accepted by Gemini. | Update the model list, shorten meeting background, re-import the audio, and try again. |
| HTTP 403 / PERMISSION_DENIED | The API key, Google project permission, or billing state does not allow this request. | Check that the key is correct and not blocked, then review Google AI Studio or Google Cloud project permissions. |
| HTTP 404 / NOT_FOUND | The selected model or an uploaded Gemini file resource could not be found. | Update the model list in Settings or retry the audio processing after a short wait. |
| HTTP 429 / RESOURCE_EXHAUSTED | Your Gemini quota or rate limit was reached. | Wait for the in-app retry countdown, switch to another usable Gemini key, or review quota and billing in your Google project. |
| HTTP 500 / INTERNAL | Google encountered an unexpected error, or the input context was too long. | Retry after a short wait, reduce meeting background or input context, switch models, and check Gemini API status. |
| HTTP 503 / UNAVAILABLE | The Gemini service may be temporarily overloaded or unavailable. | Retry after a short wait, switch models, and check Gemini API status. |
| HTTP 504 / DEADLINE_EXCEEDED | Gemini could not finish within the processing deadline, often because the input is too large. | Reduce meeting background or audio input size, keep the app in the foreground, and retry. |
Official sources: Gemini API troubleshooting guide and Gemini audio understanding.
Privacy summary
Trace uses your own Gemini API key stored only in iOS Keychain. Audio and related content are sent to Google Gemini, and long-audio processing may use server-side Interactions storage that Trace attempts to clean up. See the Privacy Policy for data types, retention periods, and your choices.