Struct BridgeConfig 

pub struct BridgeConfig {
    pub openai_base: String,
    pub native_base: String,
    pub upstream_model: String,
    pub upstream_key: Option<String>,
    pub served_model: String,
    pub description: String,
    pub release_date: String,
    pub chat_template_kwargs: Value,
    pub max_input_tokens: Option<usize>,
    pub local: LocalComponents,
}

Everything Bridge::connect needs to reach one upstream model.

Fields

openai_base: String

OpenAI-compatible root, for example http://127.0.0.1:8080/v1.

native_base: String

Native root for llama.cpp’s /completion.

upstream_model: String

Model name sent upstream.

upstream_key: Option<String>

Bearer token for the upstream service, when it needs one.

served_model: String

Model id this bridge accepts from clients.

description: String

Description returned by GET /v1/models.

release_date: String

ISO release date returned by GET /v1/models.

chat_template_kwargs: Value

Template variables, for example {"enable_thinking": false}.

max_input_tokens: Option<usize>

Reject rows above this token count instead of truncating them.

local: LocalComponents

Render the chat template and/or tokenize in-process instead of asking the runtime.