Serialized API key used by this provider profile.
Provider base URL for the selected protocol.
Prompt input cost in USD per 1M cached tokens.
Capability metadata describing optional model input and output features.
Completion output cost in USD per 1M tokens.
Context token threshold where future compression should start.
Maximum context token count available to the model.
Unique persisted provider profile identifier.
Indicates whether this profile was provided by the engine.
Maximum provider/tool iterations allowed for one agent turn.
Maximum output token count used by this provider profile.
Model identifier used by this provider profile.
Human-readable provider profile name.
Prompt input cost in USD per 1M cache-miss tokens.
Provider protocol used by this profile.
Sampling temperature used by this provider profile.
Static ReadonlybuiltBuilt-in provider profiles available to every project.
Static ReadonlydeepStable UUID for the built-in DeepSeek provider profile.
Static ReadonlyopenStable UUID for the built-in OpenAI provider profile.
StaticcloneCreates a detached copy of a provider profile.
Persisted configuration for one LLM provider endpoint.