Bring any AI model
Connect OpenAI, Anthropic, Google Gemini, xAI, AWS Bedrock, or run your own models with Ollama and llama.cpp — switch or add providers anytime without rewriting your applications.
One AI gateway. Every model. Your infrastructure.
WaddleAI lets your teams use the AI models they want without handing your data or budget to a single vendor. See exactly what each team and project is spending, and keep prompts and outputs on infrastructure you control — plus a built-in AI coding assistant for your developers.
Highlights
See exactly what you’re spending
Track usage and cost by team, project, and individual key, with spending limits that kick in automatically before a budget is blown.
Keep your data on your own infrastructure
Run AI models entirely on hardware you control when data can’t leave the building, with sensitive content automatically screened before it goes anywhere.
Works with the tools you already use
Point the OpenAI or Anthropic tools your teams already rely on at WaddleAI — no code changes required.
One provider’s bad day doesn’t become yours
If an AI provider goes down or slows to a crawl, WaddleAI automatically routes around it so your team keeps working.
An AI coding assistant, included
Give developers a coding assistant that runs on your own infrastructure or routes through the platform — in the CLI, VS Code, or any MCP-compatible tool.
Built for real deployments
WaddleAI is released under a Limited AGPL-3.0 license with commercial-use restrictions and a Contributor Employer Exception — free for personal and internal use, source-available for review.
Kubernetes via Helm is the only supported deployment path, so it runs the same way in a homelab, a datacenter, or a cloud cluster you already operate.
7AI providers supported
2popular AI APIs it plugs into out of the box
4built-in access levels for your team
See it in action
