ServicesModelsPricingUse CasesRegionsOverview
Features

Sovereign AI, turnkey

From a simple API to fine-tuned models, everything stays in Canada.

💡

Smart Routing

Automatic routing to the optimal model (Lite → Standard → Pro) based on complexity.

🔌

OpenAI-compatible API

Change one line of code, not your stack. Python, Node.js, curl SDK — everything works.

📚

Built-in RAG

Native vector database to enrich responses with your data. 1 GB free.

⚙️

Fine-tuning

Train a custom model on your data. Automatic deployment.

🍁

Sovereignty

Data 100% in Canada, compliant with Law 25 and PIPEDA. Zero international transfer.

Model Catalog

Pre-optimized open-source models

Available on the Inference API and for download on your dedicated GPU instances.

Llama 3.1

8B · 70B — Meta

Mistral

7B · Mixtral 8x7B

Qwen 2.5

7B · 32B · 72B

CodeLlama

7B · 34B — Code

Gemma 2

9B · 27B — Google

Embeddings

BGE · E5 · Nomic

Catalog constantly expanding. You can also deploy any Hugging Face-compatible model on your dedicated GPU instances.

Pricing

Per-token billing

All prices in CAD, before taxes (GST/QST).

JRAF Models

ModelInput (/1M tokens)Output (/1M tokens)Use case
JRAF Lite0.35 $1.40 $Chatbot, classification, triage
JRAF Standard Popular1.50 $6.00 $Writing, analysis, code, RAG
JRAF Pro8.00 $32.00 $Complex reasoning, agents

Complementary services

ComponentPrice (CAD)
JRAF API accessFree
Embeddings0.05 $ per 1M tokens
Vector database (RAG) — 1 GBFree
Additional vector storage0.10 $/GB/month
Fine-tuning (training)15.00 $ per 1M tokens
Fine-tuned model hosting2.00 $/h (active) · 0 $ (inactive)
Starter credit5.00 $ free

Example: JRAF agent, 100 conversations/day (80% Lite, 20% Standard) ≈ $13/month. Compliant with Law 25 and PIPEDA.

Use Cases

Sovereign AI for your business

Real use cases where Canadian hosting makes all the difference.

🏥 Healthcare & Sensitive Data

Analyze patient records, summarize medical reports, assist with diagnosis — with the guarantee that data stays in Canada and complies with provincial laws.

⚖️ Legal Services

Case law research, contract analysis, assisted drafting. Your clients' confidential data protected by Canadian sovereignty.

🏦 Financial Services

Risk analysis, fraud detection, intelligent customer service. Canadian regulatory compliance ensured by data localization.

🏛️ Public Sector

Automation of citizen services, document processing, policy analysis. 100% Canadian infrastructure for government requirements.

💬 Chatbots & Assistants

Deploy AI assistants on the Inference API with the OpenAI-compatible API. Migrate from OpenAI/Azure by changing a single URL.

🔍 RAG & Semantic Search

Combine UBQT embeddings with your vector databases for intelligent search over your internal documents.

Regions

AI Compute in Canada

Your models run exclusively on Canadian infrastructure.

Quebec Available at Launch

CAN-QC01

Quebec Available at Launch

CAN-QC02

Ontario Planned

CAN-ON01
Overview

Zero to Inference in 2 Minutes

Here is what the Inference API experience looks like.

1

Get Your API Key

One key for all models on the Inference API.

$ ubqt ullm apikey create --name mon-app
Key created — ubqt-sk-xxxxxxxxxxxx
2

Send Your First Request

OpenAI-compatible API — use the SDK you already know.

# Compatible with the OpenAI SDK
$ curl https://api.can-qc01.ubqt.cloud/v1/chat/completions \
  -H "Authorization: Bearer ubqt-sk-xxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "jraf-standard",
    "messages": [{
      "role": "user",
      "content": "Bonjour depuis le Cloud d'ici!"
    }]
  }'
3

Or With the Python SDK

Change the base_url — the rest of your code stays the same.

# pip install openai
from openai import OpenAI

client = OpenAI(
  base_url="https://api.can-qc01.ubqt.cloud/v1",
  api_key="ubqt-sk-xxxx"
)

response = client.chat.completions.create(
  model="jraf-standard",
  messages=[{"role": "user", "content": "Bonjour!"}]
)

print(response.choices[0].message.content)
# Hello! How can I help you today?

Sovereign AI is coming

Sign up to be notified when LLM / AI services become available.

We will only contact you for the launch. No spam.