Changelog
What's shipping
Product updates across the platform — what's live, what's in beta, and what's coming next.
Per-hour metering for compute & GPU
Virtual machines and GPU pods are now metered hourly against your credit balance, with a 7-day grace window and clear usage breakdowns in the dashboard.
Custom OS images for VPS
Bring your own images by URL import or snapshot, then launch new instances from them in any region. Billed per GB-month of stored image data.
Managed vector collections (pgvector)
Spin up managed vector collections on Postgres with hosted embedding models — build RAG and semantic search without running your own vector store.
Fine-tuning with auto-deploy
Train LoRA adapters on open models and deploy the result to a dedicated endpoint in one flow. We're expanding base-model coverage during beta.
Dedicated model hosting
Deploy your own weights or a HuggingFace model to a dedicated GPU endpoint with per-hour pricing and idle auto-stop. Try it and send us feedback.
Domain registration & marketplace
Search, register, transfer, and manage domains directly from the dashboard, with DNS management and a domain marketplace built in.
Reserved GPU clusters (self-serve)
Reserve committed B300 / B200 / H200 capacity with discounted pricing directly from the dashboard — no sales call required.
Batch inference API
Submit large asynchronous inference jobs and retrieve results at a lower price point — ideal for offline enrichment and evaluation workloads.
Team workspaces & role-based access
Invite teammates, share projects, and control access with granular roles across all services.