Usage-based pay-as-you-go pricing with hourly compute charges; committed contracts available for volume discounts. Hosted option with Anyscale-managed infrastructure or Bring Your Own Cloud deployment available.
Free: $0, includes up to 5 members, 100 context nodes, and 100 MB of indexed content. Team: $25 per seat per month (or $200 per seat billed annually), includes up to 25 members, 1,000 context nodes, and 1 GB of indexed content. Enterprise: custom pricing, includes Forward Deployed Engineers, Hybrid
Key features
Multimodal data curation at scale
Distributed model training with elastic scaling
Batch embedding generation at scale
Multi-cloud deployment and orchestration
Cluster-backed development environments
Workload-specific observability and debugging
Production-grade managed Ray clusters
Access control and governance
GPU budget controls and cost attribution
Post-training with SkyRL and veRL support
Model-agnostic: runs on Claude, GPT, Gemini, or open-weight models
Full observability: trace every tool call, retrieval, and reasoning step
Isolated execution: tool calls run in isolated environments by default
Code mode: reduces token usage to one-third of traditional approaches
Self-hosted on your infrastructure: cloud, on-prem, or air-gapped
MCP server integration: works with existing MCP servers unchanged
Open source under MIT license
Automatic model failover and switching
What makes it different
Built by creators of Ray, the world's most widely adopted AI compute engine
Multi-cloud deployment without code changes
Feels local but runs distributed
Unified GPU pooling across clouds and regions
Isolated by default with no direct system access
Reduces token usage to approximately one-third
Runs end-to-end air-gapped with no external network calls
Full transparency through open agent-observability spec (AVP)
No vendor lock-in: switch models without rewriting agents