Module provider 

Model provider abstraction — pluggable LLM backends.

Built-in backends: DeepSeekProvider. The default ModelProvider::stream_complete wraps ModelProvider::complete so non-streaming providers automatically get a (single-chunk) streaming impl.

Re-exports

pub use deepseek::DeepSeekBalance;

pub use deepseek::DeepSeekBalanceInfo;

pub use deepseek::DeepSeekChatOptions;

pub use deepseek::DeepSeekConfig;

pub use deepseek::DeepSeekFimChoice;

pub use deepseek::DeepSeekFimRequest;

pub use deepseek::DeepSeekFimResponse;

pub use deepseek::DeepSeekModel;

pub use deepseek::DeepSeekModelList;

pub use deepseek::DeepSeekProvider;

pub use deepseek::DeepSeekResponseFormat;

pub use routed::RoutedModelConfig;

pub use routed::RoutedProvider;

Modules

deepseek

DeepSeek API provider.

routed

Hint-based model routing provider.

Structs

CompletionRequest

All inputs a provider needs to generate a single completion.

CompletionResponse

Aggregated result of a non-streaming completion.

ErasedProvider

A newtype that implements ModelProvider by delegating to an erased &dyn ModelProvider reference. This is useful when adapting an existing borrowed provider to APIs that require a concrete ModelProvider type.

TokenUsage

Token accounting reported by the provider.

Enums

ModelHint

Semantic routing hint — describes the nature of a provider call so that a routing provider can select an appropriate model.

ProviderEvent

One unit of incremental output from a streaming completion.

StopReason

Why the model stopped emitting tokens.

Traits

ModelProvider

Abstract LLM backend. Implement this to plug in a new model provider.