Struct OpenAICompatible 

pub struct OpenAICompatible { /* private fields */ }

Available on crate feature openai only.

Shared OpenAI-compatible client implementation.

Implementations

impl OpenAICompatible

pub fn new(config: OpenAICompatibleConfig) -> Result<OpenAICompatible, AdkError>

Available on crate feature models only.

Create a new OpenAI-compatible client.

pub fn new_with_reasoning_effort( config: OpenAICompatibleConfig, reasoning_effort: Option<OpenAIReasoningEffort>, ) -> Result<OpenAICompatible, AdkError>

Available on crate feature models only.

Create an OpenAI-compatible client with the complete OpenAI reasoning vocabulary.

This supports newer values such as none, xhigh, and max without changing the backward-compatible OpenAICompatibleConfig::reasoning_effort field type.

pub fn with_default_headers( self, headers: HeaderMap, ) -> Result<OpenAICompatible, AdkError>

Available on crate feature models only.

Adds default HTTP headers to every request, replacing earlier defaults with the same name.

The bearer token and organization headers set per request take precedence, and redirects stay disabled. Unlike builder methods that only store a value, this method rebuilds the underlying reqwest client, so it returns Result.

Example

use adk_model::{OpenAICompatible, OpenAICompatibleConfig};
use http::header::{HeaderMap, HeaderValue};

let mut headers = HeaderMap::new();
headers.insert("x-session-id", HeaderValue::from_static("conversation-1"));
let client = OpenAICompatible::new(OpenAICompatibleConfig::new("api-key", "model"))?
    .with_default_headers(headers)?;

Errors

Returns an Internal error with code model.openai.http_client_init when the HTTP client cannot be built.

pub fn with_retry_config(self, retry_config: RetryConfig) -> OpenAICompatible

Available on crate feature models only.

Set the retry configuration (builder pattern).

pub fn with_request_adapter( self, adapter: Arc<dyn Fn(&mut Value, &mut HeaderMap) -> Result<(), AdkError> + Sync + Send>, ) -> OpenAICompatible

Available on crate feature models only.

Customize generation request fields and headers before each HTTP attempt.

The adapter runs under this client’s RetryConfig; see RequestAdapter for an example.

pub fn set_retry_config(&mut self, retry_config: RetryConfig)

Available on crate feature models only.

Set the retry configuration (mutable reference).

pub fn retry_config(&self) -> &RetryConfig

Available on crate feature models only.

Returns the current retry configuration.

pub fn with_reasoning_replay(self, enabled: bool) -> OpenAICompatible

Available on crate feature models only.

Enables or disables assistant reasoning replay through reasoning_content.

Disabled by default. Passing true selects reasoning_content, replacing any previous field selection; passing false clears the selection. This controls history serialization, not whether the model thinks.

Example

use adk_model::{OpenAICompatible, OpenAICompatibleConfig};

let client = OpenAICompatible::new(OpenAICompatibleConfig::new("key", "model"))?
    .with_reasoning_replay(true);

pub fn with_reasoning_replay_field( self, field: ReasoningReplayField, ) -> OpenAICompatible

Available on crate feature models only.

Enables assistant reasoning replay with an explicit provider field.

Newer vLLM servers use ReasoningReplayField::Reasoning. Older deployments may still require reasoning_content; select the field the endpoint accepts. The last call to either reasoning-replay builder determines the selection.

Example

use adk_model::{
    OpenAICompatible, OpenAICompatibleConfig, ReasoningReplayField,
};

let client = OpenAICompatible::new(OpenAICompatibleConfig::new("key", "model"))?
    .with_reasoning_replay_field(ReasoningReplayField::Reasoning);

Trait Implementations

impl Llm for OpenAICompatible

fn name(&self) -> &str

Returns the model identifier (e.g., “gemini-2.5-flash”).

fn generate_content<'life0, 'async_trait>( &'life0 self, request: LlmRequest, stream: bool, ) -> Pin<Box<dyn Future<Output = Result<Pin<Box<dyn Stream<Item = Result<LlmResponse, AdkError>> + Send>>, AdkError>> + Send + 'async_trait>>

where 'life0: 'async_trait, OpenAICompatible: 'async_trait,

Generates content from the given request, optionally streaming.

fn schema_adapter(&self) -> &dyn SchemaAdapter

Returns the schema adapter for this provider.

fn uses_interactions_api(&self) -> bool

Returns true if this model is configured to use a server-managed environment (e.g., Gemini Interactions API) where the provider owns the tool-calling loop and filesystem.