DGX Cloud Lepton Free API and NVIDIA Build Migration Guide
🌍 International ✅ Free
Lepton AI was founded by Yangqing Jia (PyTorch co-creator) and later moved into the NVIDIA DGX Cloud / Build path. For developers, the key question is not whether the old lepton.ai endpoint works, but whether the current NVIDIA console exposes testable models, credits, RPM, endpoints, and pricing. Compare it with NVIDIA Build, Groq, Fireworks, and Together before production.
🎁 Free Tier
Daily Limit: 10M tokens/day
| Model | Context | Limit | Notes |
|---|---|---|---|
| Llama 3.3 70B | 128k | 10 RPM | Free to use, built by Chinese-American team |
| Mixtral 8x7B | 32k | 10 RPM | Available in free tier |
🔑 Free API
Free Credits: $10 free credits
Rate Limit: 10 RPM
Historical record shows $10 new-user credits; the service moved into the NVIDIA DGX Cloud / Build path, so current credits, model catalog, and endpoints must be verified in the NVIDIA console.