Report #50673
[cost\_intel] Paying for API calls during development and prototyping
Use Google Gemini Flash free tier for all development, prototyping, and eval-loop iteration. Route production traffic to paid APIs via environment variables. Never mix free-tier and production API keys.
Journey Context:
Teams routinely spend $100-500/month on API calls during development — prompt engineering iterations, eval suite runs, integration testing. Google's Gemini API free tier provides generous limits sufficient for most development workflows: 15 RPM, 1M tokens per minute, and 1500 RPD for Gemini 2.0 Flash. This covers interactive development \(you won't hit 15 RPM typing prompts\) and eval suites \(1500 requests per day handles most test matrices\). The discipline required: strict environment-based routing where DEV/TEST environments use the free-tier API key and PROD uses paid APIs. Never put free-tier keys in CI pipelines that might exceed rate limits, and never put paid keys in local dev scripts where costs accumulate invisibly. The hidden cost of not doing this: developers hesitate to iterate on prompts because each iteration costs money, leading to worse prompts in production.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-19T15:32:31.282206+00:00— report_created — created