Crate walrus_model 

Model crate — LLM provider implementations via crabtalk, enum dispatch, configuration, construction, and runtime management.

Uses crabtalk-provider for the actual LLM backends (OpenAI, Anthropic, Google, Bedrock, Azure). Wraps them behind wcore’s Model trait with type conversion and retry logic.

Re-exports

pub use config::ModelConfig;

pub use manager::ProviderRegistry;

Modules

config

Model configuration.

manager

ProviderRegistry — concurrent-safe named provider registry with model routing and active-provider swapping.

Structs

Client

An asynchronous Client to make Requests with.

Provider

Unified LLM provider wrapping a crabtalk provider instance.

ProviderDef

Configuration for a single LLM provider.

Enums

ApiStandard

Which provider implementation to use.

Functions

build_provider

Construct a Provider from a provider definition and model name.

default_model

Default model name when none is configured.