Koyeb

Koyeb is a next-generation serverless platform that lets developers deploy and scale applications, containers, and ML models globally without managing infrastructure. It provides instant API endpoints, native autoscaling (including scale-to-zero), and access to GPUs and accelerators across 50+ locations. Koyeb supports Git-based and CLI deployments, managed Postgres with pgvector, and pay-per-use pricing with a forever free tier.

• Serverless containers — Deploy production-grade containers with zero configuration using a Git push or the CLI; Koyeb provisions and scales instances automatically.
• GPU & accelerator support — Run models and inference on NVIDIA, AMD, Intel, and other accelerators; choose instance types and pay per second for compute.
• Global edge locations — Deploy to 50+ regions to reduce latency and improve availability for worldwide users.
• Autoscaling & scale-to-zero — Use built-in autoscaling for high throughput and negligible cold starts, and scale to zero to avoid charges when idle.
• Managed Postgres + pgvector — Store, index, and search embeddings with a fully managed serverless Postgres instance.
• Instant API endpoints & logs — Deploy an API endpoint in seconds and troubleshoot with real-time logs and instance access.
• Pay-per-use pricing & free tier — Start for free with the Hobby plan; upgrade to Pro or Scale for included compute credits and priority support.

To get started, sign up on the web console, connect a Git repository or use the CLI to push your app, pick instance hardware (CPU/GPU), and deploy — Koyeb handles routing, autoscaling, and global distribution.