What It Is
Every property in the lab needs inference, and 4ort.io is where it comes from: an OpenAI-compatible multi-provider LLM gateway. Anything that speaks the standard API can point at it and get capability-based routing across local models and cloud providers. The gateway picks the right brain for the job instead of forcing every caller to hardcode one.
Local inference runs on an ik_llama.cpp cluster; cloud reach comes via OpenRouter. The result is one endpoint that abstracts the whole spectrum, from self-hosted models to frontier providers.
Then there's the signature move: "4ort AI", the Council. Ask one question through the model id 4ort/council and the gateway fans it out to a panel of frontier models, synthesizes their collective answer, and delivers it in the lab's voice. One endpoint, many brains.
Highlights
- OpenAI-compatible gateway: any standard client can point at it, no rewrites
- Capability-based routing across local models and cloud providers
- The Council: fan a question out to a panel of frontier models and synthesize their collective answer
- One model id: the whole Council behind
4ort/council, delivered in the lab's voice - Local + cloud spectrum: an ik_llama.cpp local cluster alongside OpenRouter reach
- One endpoint, many brains: the inference layer the rest of the lab builds on
Status
● Live.