Starter
- Rack-mount 1U appliance
- Up to 20 concurrent users
- 13B instruct + embedding + vision
- Full RAG, OPA policy, virtual keys
- 99.5% uptime SLA
- Priority fleet patches & rollback
- Merkle-signed audit with offline verify
- Phone + chat support, 8×5
Hardware, the software, the fleet service, the audit pipeline, and the portal are all in one monthly price. Leases are 36 months and include ownership transfer at the end of term.
You pay with flexible, enterprise-ready models. Pre and post-paid, with a model that fits your needs.
All tiers include fleet management, security patches, model updates, and audit infrastructure. Hardware ownership transfers after 36 months. Trial units available for qualified prospects.
Every plan includes a fixed monthly fee covering the appliance, fleet management, security patches, and support. On top of that, you pay per-token based on actual gateway usage — so you only pay for what your teams consume. Both pre-paid and post-paid billing modes are available.
All tiers include access to leading models from Anthropic, OpenAI, DeepSeek, and xAI through the gateway. Model routing, failover, and load balancing are handled automatically. You can configure which models each team or virtual key has access to.
Not at launch. We need a specific silicon and firmware baseline to honour our SLAs, and every appliance on the fleet shares the same OS image. Customers who already have qualifying hardware should talk to us — we have a path for it in 2027.
Monitoring, incident response, OS and model updates, remote diagnostics, hardware replacement, and policy authoring help. The difference between tiers is response time and dedicated engineering. All tiers include Merkle-signed audit trails and the operator portal.
All appliance-side data — prompts, documents, audit bodies — never leaves your building. The central plane stores only metadata, and you choose its region (EU or UAE) at enrolment. Air-gapped deployment is available on Entry and Premium tiers.
Each team or application gets its own virtual API key with configurable rate limits, model access, and spend caps. Usage is metered per key, giving you full visibility and cost control across departments — without sharing upstream provider credentials.
Spend 20 minutes with one of our deployment engineers. We'll walk through your workload, pick the right tier, and ship an appliance to your office within two weeks.