Free to test. Free to own.
Test and iterate on your AI chatbots instantly on our free managed infrastructure. Once you've proved the value, deploy the open-source code to your own AWS VPC.
Managed Trial
Instantly build and test your agents over our hosted infrastructure.
- Up to 250,000 requests/mo
- 100MB Vector Storage
- Zero configuration required
- Instant Next.js dashboard access
Self-Host (OSS)
Own the code. Maintain absolute privacy by deploying to your own AWS.
- Unlimited Agents & Requests
- Total Data Privacy (Your VPC)
- Extensible Python backend
- 100% MIT Licensed Codebase
Real AWS Running Costs
While VegaRAG itself is free, hosting it on AWS does incur infrastructure costs. Here's what you can expect to pay Amazon directly.
Fargate + ALB
Compute Overhead
~ $25 to $40 / month
Runs the Next.js React frontend and the FastAPI Python backend asynchronously on serverless containers via an Application Load Balancer.
Pinecone Serverless
Vector Database
Pay per Read/Write usage
Pinecone offers $100 in free serverless credits. Extremely cost-effective for multi-tenant setups via namespace isolation compared to indexed endpoints.
Bedrock Nova
Token Generation
$0.035 per 1M Output Tokens
Amazon Nova Micro is ridiculously cheap. You will likely pay pennies per month for all your RAG streaming needs.
S3 & DynamoDB
Core Storage
Practically free under Free Tier
Stores uploaded PDFs, Excel sheets, and fast metadata for user bots. Usually remains entirely within the AWS Free Tier limitations.