HFAI Cloud · integration
Continue
on HFAI Cloud.
One YAML block, four keys that matter.
Four keys in a model entry.
Its own YAML shape,
our endpoint inside it.
Continue's own documentation introduces it as an open source AI code assistant for VS Code and JetBrains, built so that developers can create, share and use custom agents. The page it keeps for OpenAI models documents a YAML model entry whose keys matter here: the provider name, the model identifier, an apiBase override for endpoints that are only OpenAI-compatible, and the apiKey that the entry authenticates with. Checked against docs.continue.dev.
name: HFAI Cloud
version: 0.0.1
schema: v1
models:
- name: GPT-OSS 120B
provider: openai
model: gpt-oss:120b
apiBase: https://api.heartflowai.net/v1
apiKey: hfai_live_…
The entry above is the shape Continue documents, with the four values that belong to this endpoint. Because the key and the base URL sit on the entry, a second entry with another model identifier is all it takes to add another model to the picker, and the same assistant then answers from whichever one you choose.
Continue is developed by its own team and distributed under its own terms. HeartFlowAI does not maintain, sponsor or endorse Continue, and no part of this page was written or reviewed by its maintainers. The configuration keys are Continue's, checked at docs.continue.dev/customize/model-providers/top-level/openai.
The same four keys, another model.
Copy the entry,
change the identifier.
Continue keeps the endpoint and the key on the model entry, so a catalogue with several models becomes several entries in the same file rather than several configs.
The chat-completions shape is documented in the API reference, the balance side on HFAI Cloud, and GPT-OSS 120B needs no balance at all.
Two assistants, one key.
Continue holds its models in a file you own, which makes the endpoint and the key portable between machines without exporting anything from an app.