the provider catalog

native

{"rooms":[{"slug":"qwen-qwen3-8b","id":"Qwen/Qwen3-8B","aliases":["qwen"]}],"market":{"id":"native","name":"Native Inference","requires_key":false,"api_dialect":"openai-chat","tags":["local","open-source"],"data_policy":{"trains":"no","retention":"none","zdr":"local","source":"local runtime · no egress by construction"},"models":[{"alias":"qwen","model":"Qwen/Qwen3-8B"}]}}

A vendor the released binary can reach. Native Inference is in the engine's catalog, so a workflow can point a task at it today · it is not one of the 17 providers the SPEC names, which is the stronger promise: those carry a conformance contract and a room written against it.

the wireopenai-chat · no key needed · [object Object]

what it doeslocal · open-source

the seats it serves

1 seat

The aliases a file may write, and the exact upstream id each one pins. A model with its own room in the released catalog links to it.

native/qwenQwen/Qwen3-8B

Every provider the binary knows Every model, with its price