May 2025
31st May 2025: Runtime Flexibility, Faster Remote Run, and Hugging Face Improvements
Latest Enhancements & Features:
- Runtime Optimization: We’ve removed region restrictions from runtimes. Now, developers can use any runtime in any region—bringing runtime configs closer to the code and giving you more flexibility during deployment.
- 🤗 Hugging Face Import Improvements: To reduce model import failures, we’ve added validation checks—especially for cases where access to model weights is restricted. We’ve also introduced vLLM plugin support for faster inference with Hugging Face models.
- Remote Run – Speed Enhancements: HF models used in Remote Run now benefit from caching, making subsequent executions significantly faster—ideal for quick testing and iteration cycles.
- Streaming Logs for Model Import You can now view logs in real-time during model imports, helping you track build progress more accurately and debug faster.