One-click apps and models

Open-weight mixture-of-experts model for high-throughput inference.

Multimodal open-weight model for text and vision applications.

Speech-to-text transcription and translation across 90+ languages.

State-of-the-art open image generation served on regional GPUs.

High-throughput serving engine with paged attention, ready to deploy.

Self-hosted chat interface wired directly to your Edge Node endpoint.

S3-compatible object storage for datasets, artifacts, and model files.

Interactive notebooks with GPU access for experimentation and evals.

Lightweight self-hosted Git service for private model and prompt repos.

OpenAI-compatible gateway for routing, budgets, and key management.

Relational database with vector search for RAG and embeddings.

In-memory data store for caching, queues, and semantic caching.

Purpose-built vector database for similarity search at scale.

Columnar analytics database for logging tokens, traces, and events.

The world's most popular open-source relational database.

Document database with flexible schemas for application data.

Dashboards for token throughput, latency, and spend across your keys.

LLM tracing and evaluation for every request through your gateway.

Metrics collection and alerting for your deployed services.

Self-hosted status pages and endpoint monitoring with alerts.

Workflow automation with native LLM nodes pointed at your endpoint.

Orchestrate batch inference, evals, and data pipelines on a schedule.

Durable execution for long-running, fault-tolerant agent workflows.

Centralized secrets management for API keys and credentials.

Identity and access management with SSO for your internal tools.

Modern reverse proxy with automatic TLS in front of your services.

Fast, modern VPN for private access to your inference endpoints.