Spice AI
A SQL and hybrid search platform for governed enterprise data access by applications and AI agents, available in self-hosted and managed deployments.
About Spice AI
Spice AI is a data platform for applications and AI agents that need fast access to operational, analytical, streaming, and object-store data. Its open-source engine combines SQL queries with hybrid search and can run as a sidecar, an independent microservice, a tiered deployment, or a managed cloud service. Teams can use temporary, policy-scoped datasets to limit what models and agents can reach, while built-in task history records trace IDs, timings, and labels for queries and AI completions. The platform exposes HTTP, Arrow Flight, Arrow Flight SQL, and ADBC APIs, with SDKs for Python, Go, Node.js, Rust, C#, and Java. Teams must choose between self-hosted community operation, the managed cloud plans, and enterprise licensing. Cloud limits differ materially by plan: instance size, storage, query concurrency and timeout, app count, commercial licensing, support, availability, and SLA all change. Security controls include RBAC, encryption in transit and at rest, enforced MFA, least-privilege access, and SOC 2 Type II compliance.
Best for
Best use cases
Serving governed enterprise context to AI agents
Accelerating queries across operational and analytical data
Monitoring query and completion behavior with distributed traces
Compare before choosing
Choose Community for Apache 2.0 self-hosting and community support
Choose Spice Cloud for a managed service with plan-based capacity
Choose Enterprise for dedicated infrastructure, 24/7 support, and a 99.9+ SLA
Features
Governed agent data access
Provision temporary, scoped datasets so each model or agent receives only the data needed for its work.
SQL and hybrid search
Query operational, analytical, streaming, and object-store data through a shared engine for applications.
Operational tracing
Inspect trace IDs, timings, labels, query failures, and AI completion steps through built-in task history.
Pros
- The Apache 2.0 community edition supports self-hosting without a commercial license fee.
- Multiple deployment patterns let teams place the engine beside an application, in a service, or in managed cloud infrastructure.
- RBAC, encryption, MFA, scoped datasets, and task history address practical governance and audit needs.
Cons
- Paid cloud plans impose different query concurrency, timeout, storage, and app limits, so capacity planning is required.
- The Developer cloud plan is limited to one user and five apps, which may be restrictive for a team.
- Dedicated infrastructure, multi-region availability, 24/7 support, and the highest SLA require custom-priced Enterprise service.
Pricing Plans
Developer
- Single instance with 2 vCPU and 4 GB memory
- Up to 16 concurrent queries with a 90-second timeout
- One user and five apps
- Community support
Pro for Teams
- Single instance with 4 vCPU and 8 GB memory
- Up to 64 concurrent queries with a five-minute timeout
- Unlimited users and ten apps
- Standard support through private Slack and email
Enterprise
- Multi-instance deployment with custom CPU and memory
- Up to 1024 concurrent queries with a 30-minute timeout
- Multi-cloud, multi-region high availability
- Premium 24/7 on-call support
* Pricing information is for reference only. Please refer to the latest announcement on Spice AI's official website.