The free tier is roughly 40 requests/minute, but popular models (like DeepSeek Pro) can hit rate limits sooner, and there are per-model and daily caps. It's enough for normal development sessions. If you hit limits often, you can create multiple free accounts and rotate keys, or switch to a less-busy model.
Frequently asked questions
Answers to common concerns about setup cost, model behavior, and operational reliability.
Security Reminder
- ⚠️ Never paste your
nvapi-...key in any chat or public forum - ⚠️ Never commit shell profile files or config files with keys to GitHub
- ✅ Use environment variables (
os.environ/NVIDIA_NIM_API_KEY) - ✅ Regenerate key immediately if accidentally exposed