One license. Every surface.
Each Curie license includes the chat surface, the AI gateway for your apps, and the
document-processing pipeline.
Tiers differ on scale, governance depth, and customization — not on which features
you get.
LAUNCH-YEAR Special: 35% Off All Plans!
Limited time offer automatically applied to all pricing below
Basic
Secure team chat & document processing for a single org
$23,500/year
$15,275
/year
Save $8,225 with LAUNCH-YEAR
Offline deployment
- Up to 50 users
- Single organization
- Secure team chat with collaborative workspaces
- Shared context items & discussion threads
- Document pipeline: PDF → markdown with local OCR
- Text-query batches over uploaded corpora
- Per-user cost & rate limits, basic admin controls
- Standard email support
Most Popular
Advanced
Everything in Basic + the AI gateway for your apps
$37,200/year
$24,180
/year
Save $13,020 with LAUNCH-YEAR
Offline deployment, full integration surface
- Everything in Basic, plus:
- Up to 200 users total
- Up to 3 organizations
- AI Gateway: programmatic access for in-house apps NEW
- Tool forwarding to OpenAI / Anthropic / Azure
- Read-only audit of programmatic conversations
- Agent profiles with layered prompt stacks
- Usage groups with capability overrides
- SSO / SAML integration
- Premium deployment support
Ultimate
Everything + compliance, audit, and customization
$83,150/year
$54,048
/year
Save $29,102 with LAUNCH-YEAR
Full enterprise capabilities
- Everything in Advanced, plus:
- Unlimited users
- Unlimited organizations
- Compliance-grade audit logging & retention
- Higher document-pipeline throughput & vision-page caps
- Per-key API rate & cost overrides
- White-label customization
- Custom integrations & priority feature requests
Feature comparison
| Basic | Advanced | Ultimate | |
|---|---|---|---|
| Secure team chat | |||
| Collaborative workspaces & discussion threads | |||
| Shared context items (uploads, citations) | |||
| Agent profiles (layered prompt stacks) | Single profile | ||
| AI gateway for your apps | |||
| Programmatic API (HTTP + SSE) | — | ||
| API key management & rotation | — | ||
| Tool forwarding (OpenAI / Anthropic / Azure) | — | ||
| Per-key rate & cost overrides | — | Per-org defaults | |
| Document pipeline | |||
| PDF → markdown with chunked vision | |||
| Local Tesseract OCR pre-pass & auto-rotate | |||
| Focus zones & prefer-literal mode | — | ||
| Throughput & vision-page caps | Standard | Standard | Tunable |
| Governance & deployment | |||
| Per-user cost & rate limits | |||
| Per-org capability gates | |||
| Usage groups with capability overrides | — | ||
| SSO / SAML | — | ||
| Compliance-grade audit logging & retention | — | — | |
| White-label customization | — | — | |
| Offline / air-gapped deployment | |||
Frequently Asked Questions
Curie is a self-hosted control plane that runs on your infrastructure. Pricing is per-deployment because the platform value is the same whether you have ten chat users, a hundred, or a fleet of internal apps calling the gateway. You bring your own LLM credentials, so the model spend goes directly to OpenAI / Anthropic / Azure under your existing contracts — we don't mark up tokens.
The AI gateway is a programmatic API (HTTP + SSE) that lets your in-house apps, agents, and batch jobs talk to LLMs through Curie instead of holding vendor keys themselves. It enforces the same rate, cost, and capability limits as the chat surface, forwards tool catalogs to the underlying provider, and persists every conversation for audit. It's included in the Advanced and Ultimate plans. See the API page for wire-shape details.
No. The OCR pre-pass runs locally with Tesseract; PDFs are rasterized with Poppler / ImageMagick on the same box Curie runs on. The only external call user content can trigger is to the LLM provider your operator configured — never Google Vision, AWS Textract, or Azure Document Intelligence (unless that Document Intelligence instance is inside your own tenant and you wire it up explicitly).
Curie supports on-premises, private cloud (your VPC), restricted-cloud tenants such as Azure GCC-High, and fully air-gapped environments. All plans include deployment support; Advanced and Ultimate include hands-on premium deployment assistance.
Yes. Upgrades take effect immediately; downgrades process at your next renewal period. Contact your account manager or support team to discuss plan changes.
OpenAI, Anthropic Claude, Azure OpenAI, and any custom OpenAI-compatible endpoint. You configure providers per-organization with your own credentials. The same model catalog is exposed both to chat users and to apps calling the AI gateway, with capability gates that let you allow some providers for chat-only or API-only.
Ultimate-tier white-label options cover branding, logos, colors, and domain to match your organization's identity — useful if you want to present Curie as an in-house tool to staff or downstream tenants.
Once your account is created and you've paid for your plan, you'll receive two things: a license key and read access to a private container registry. Pull the image, install the license, and you're running. Most clients pull images directly to a staging server or sync them into an internal registry on a schedule. Test in staging and back up data before promoting to production. Full operator documentation is available in your account.
Have questions or need a custom solution?
Our team is here to help you find the right plan for your organization.
Contact Sales