Every model. Every cloud. No lock-in.
Cana is model- and cloud-neutral by design. Route each step to the best model for the job, and run wherever your data and regulators need you to.
The world’s best models, routed by task.
Anthropic
AvailableClaude family
Frontier reasoning, coding and agentic tool use
OpenAI
AvailableGPT family · Codex
General reasoning, coding and multimodal
xAI
AvailableGrok family
Frontier reasoning with real-time knowledge
Moonshot AI
AvailableKimi
Long-context, efficient agentic coding
Xiaomi
AvailableMiMo
Efficient open models for high-volume work
OpenRouter
AvailableHundreds of models
One key for a long tail of open and specialty models
Gemini family
Multimodal and long-context reasoning
Meta
Coming soonLlama family
Open-weight models for private and sovereign deployment
NVIDIA
Coming soonNemotron & NIM
Optimized inference for enterprise and on-prem
Mistral AI
Coming soonMistral family
European open-weight and sovereign options
Your models
EnterpriseFine-tunes & private endpoints
Bring any OpenAI-compatible or custom endpoint
Model names and trademarks belong to their respective owners. Listing indicates model availability on Cana, not endorsement.
Deploy in your cloud.
AWS
Deploy Cana in your AWS account with private networking and Bedrock model access.
Google Cloud
Run on GKE with Vertex AI model access and regional data residency.
Microsoft Azure
Deploy in your Azure tenant with Entra ID and Azure-hosted models.
Oracle Cloud
Sovereign and dedicated regions for regulated industries.
GPU & accelerated compute partners.
Sovereign and on-prem SIAs need dedicated inference. We’re assembling partners across GPU clouds, silicon and in-country compute so customers can run open-weight models next to their data.
GPU cloud providers
Silicon vendors
Sovereign compute
Edge inference
Partner with Cana. Let’s build together.
Join the waitlist for early access to Super Intelligent Agents and updates as Cana opens up.