ChatVLM Enterprise
Frontier AI that never leaves your control
Every ChatVLM plan comes with contractual zero data retention and no training on your data. Enterprise goes further: run open-weight frontier models on private infrastructure — or entirely inside your own walls — with zero third-party subprocessors.
Why private models
Three degrees of verifiable control
Trust shouldn't be a promise in a policy document. Each step removes another party from the path your data takes.
- Zero data retention, everywhereContractual ZDR and no model training on every plan, from a single Pro seat to a datacenter appliance. In writing, not in a FAQ.
- Sovereign Presets in the cloudEnterprise routes sensitive work to leading open-weight models on ChatVLM-operated private infrastructure. No OpenAI, no Anthropic, no Google — zero third-party subprocessors.
- Fully air-gapped on your siteSovereign Boxes run frontier-class open models with complete network isolation. Your data never touches a network you don’t own.
Enterprise SaaS
The sovereign cloud
The same Pro and Ultra seat tiers as Teams, wrapped in the layer your security team asks for: identity, governance, and models that never touch a third party.
Enterprise
Sovereign cloudPro and Ultra seats for your whole organization, plus sovereign models, identity controls, and audit-grade compliance infrastructure.
- Everything in Teams
- Pro & Ultra seat tiers
- Sovereign Presets on private infra
- Zero third-party subprocessors
- SAML SSO + SCIM provisioning
- Workspace model allowlisting
- Custom retention & legal holds
- Immutable audit logs & SIEM export
- SOC 2 Type II & HIPAA BAA package
- Dedicated CSM & security review help
On-premises appliances
Sovereign Boxes: your models, your walls
Air-gapped appliances that run open-weight frontier models fully on-site. Unlimited seats, offline license enforcement, and quarterly model updates delivered on encrypted physical media — no network required, ever.
Sovereign Box Theta
The office-ready private AI server. High-concurrency local routing on standard wall power — no datacenter required.
From $85,000
one-time hardware+ $60,000/yr software
- 4× RTX PRO 6000 Ada — 384 GB VRAM
- Best-in-class mid-size open models
- Up to 4 models served simultaneously
- ~150 concurrent active users
- Standard 110/220 V office power
- Unlimited seats
- Quarterly offline model updates
Talk to sales
Priced by CPU/RAM configuration · Or $7,200/mo as HaaS (36-month term)
Sovereign Box Omega
Air-gap flagshipThe datacenter-class sovereign supercomputer. Runs the largest open-weight frontier models fully air-gapped.
$580,000
one-time hardware+ $150,000/yr software
- 8× NVIDIA HGX B300 — 2.1 TB HBM3e
- Frontier-class open models in memory
- Headroom for smaller routing models
- 300–500 concurrent active users
- Mission-critical SLA — 1-hour P1, 24/7
- Unlimited seats
- Secure courier-delivered updates
Talk to sales
Or $28,000/mo as HaaS · Pre-sale site survey required
Agentic AI
Private AI for your software, not just your chats
Enterprise endpoints and Sovereign Boxes speak the same standard APIs your tools already use — so coding agents, IDE assistants, and internal automations run on completely private models.
- Bring your coding agentsPoint Claude Code and other AI assistants at your private endpoint. Your source code, prompts, and diffs never leave your control — not even for the model call.
- Standard, drop-in APIsIndustry-standard API compatibility means existing SDKs, IDE plugins, and internal tools work unchanged. Swap the base URL, keep the workflow.
- Agents without the token billAgentic workloads are token-hungry. On a Sovereign Box the marginal token is free — run coding agents, CI reviews, and batch automations around the clock at flat cost.
FAQ
Frequently asked questions
Bring your most sensitive work to AI
Talk to us about Sovereign Presets, a proof-of-concept on a Sovereign Box, or a custom compliance review.