Set spend limits so a long session can't run up a bill
How to cap or watch your spending at each AI provider you use with Interviewdom.
- Updated November 19, 2026
- Easy
- About 10 min
1. Why set a limit
Interviewdom uses your own provider account, so you pay the provider directly. A bug, a forgotten session or a leaked key could use more than you expect. A hard limit stops requests when reached; an alert only emails you. Where a provider offers a hard limit, use it.
2. Provider by provider
| Provider | Type | Where |
|---|---|---|
| OpenAI | Hard limit (optional) | Organization limits → Spend → Edit spend limit → Monthly spend limit → turn on Enforce a hard limit → Save. Projects have the same under Project settings → Limits. When hit, requests fail with 429 (organization_spend_limit_exceeded / project_spend_limit_exceeded). Enforcement isn't instant, so spend can go slightly over. |
| Anthropic (Claude) | Hard limit | Spend limits → next to Monthly spend limit, Adjust limit / Set limit. When hit, requests fail with 400 "You have reached your specified API usage limits". Credits are prepaid too. |
| Google Gemini | Hard cap per project | AI Studio Spend page → Monthly spend cap → Edit spend cap. Google notes long-running tasks may go slightly over. |
| Groq | Hard limit | Settings → Billing → Limits → Add Limit → enter monthly USD → Save. API access is blocked when reached. |
| xAI (Grok) | Prepaid | Prepaid credits; by default invoiced billing is off ($0 limit), so requests stop when credits run out. |
| DeepSeek | Prepaid | Your top-up balance is the limit. |
| OpenRouter | Prepaid + per-key limit | Credits are prepaid; give the Interviewdom key its own credit limit. Over the limit returns 402. |
| Mistral | Organization limit | Spending limits can be set at organization level only (not per workspace). |
| AWS Bedrock | Alert (actions optional) | Billing and Cost Management → Budgets → create a monthly cost budget. Alerts only notify; stopping usage needs a budget action (for example attaching a Deny IAM policy). |
| Azure OpenAI | Alert only | Subscription → Budgets → Add. Microsoft: "Resources aren't affected, and your consumption isn't stopped." Cost data can lag 8–24 hours. |
| Deepgram | Free credit | $200 starting credit. |
| Ollama | Free | Runs on your PC; no bill. |
3. Extra safety tips
- Keep prepaid balances small. For prepaid providers, a $5–$10 balance is a natural cap.
- Use a separate key for Interviewdom so you can see its usage and revoke it alone.
- Set short key expirations where supported (Anthropic, AWS Bedrock, Deepgram).
- Check usage after your first interview to see what a real session costs you.
4. What Interviewdom does
Interviewdom shows a cost estimate per provider and never sends your keys anywhere except the provider. It can't see or change your provider limits; those are set in each provider's console.