Your choice of provider
Configure connections to OpenAI, Anthropic, Azure OpenAI, Amazon Bedrock, Google Vertex AI, and OpenAI-compatible endpoints.
Local hosting · Proxy · Analytics
Your models.
One gateway.
Host AI models on your hardware and proxy requests to external providers through one unified gateway. Connect your applications once, choose where requests run, and see what happens.
Routing policy controls today. A dedicated policy page is planned. Explore policy controls →
One connection. Local or external.
Your AI infrastructure,
working together.
OneAI sits between your applications and the models they use, bringing local hosting and external-provider access into a single operating view.
OpenAI- and Anthropic-compatible APIs let applications, IDEs, and agents connect through a consistent gateway. Published names, aliases, and routing determine which configured deployment handles each request.
Your applications, IDEs & agents
Model names · routing · policy controlsLocal CPU / GPUExternal providers
Local hosting · External proxying · Analytics01 / Gateway & local hosting
Make your hardware
an AI serving layer.
Load compatible GGUF models on your CPU or GPU, publish model names for clients, and manage routing alongside your external deployments.
03 / Analytics & activity
See where requests go.
Understand how they run.
Follow request activity across local and external models. Inspect routing decisions and operational signals to understand usage and investigate failures.
Policy-aware routing
Define where
requests may run.
Bring hosting, proxying, analytics, and routing policy controls together. Keep requests local, restrict eligible providers, and apply configured provider spending limits.
The models behind the gateway
Built for more than text.
Serve the model capabilities your applications need, from embedded inference to multimodal workflows.
Embedded local inference
Run compatible models directly on your own CPU or GPU, with a native engine at the heart of OneAI.
GGUF models
Load and serve compatible GGUF models, publish clear model names, and choose where they run.
Vision & images
Work with image inputs using vision models and generate images with supported image backends.
Voice
Turn speech into text and text into speech with compatible transcription and synthesis models.
Tool calling
Connect tool-capable models to application workflows through structured tool calls.
Embeddings & retrieval
Build search and retrieval workflows with embeddings, classification, and reranking models.
Capabilities depend on the model, backend, and available hardware. Configure compatible models for vision, image generation, voice, and tool calling.
Get OneAI
Your platform. Your next step.
Public binaries are coming soon. This is where you’ll find the installers, build details, and release updates.
Windows
Coming soonThe installer will be available here when the release is ready.
macOS
Coming soonThe installer will be available here when the release is ready.
Linux
Coming soonThe installer will be available here when the release is ready.
Want to explore OneAI now?
Let’s walk through local hosting, provider proxying, and your integration needs.
OneAI is proprietary software. Third-party components and model weights retain their own licenses. Asset notices.
A conversation starts here
Have something in mind?
An application to explore, a song that stayed with you, or an idea we could build together. I’d love to hear it.
Email Veera [email protected]Opens your email app.