Struct AlloyedModelProvider 

pub struct AlloyedModelProvider { /* private fields */ }

A StreamingModelProvider that cycles through multiple providers in round-robin order.

Each call to stream_response advances to the next provider in the list. This allows alternating between models with different strengths on successive turns.

Behavior

• stream_response – Uses the current provider, then advances the index.
• display_name – Returns the current provider’s name (does not advance).
• context_window – Returns the minimum context window across all providers. Returns None if any provider’s window is unknown.
• model – Returns the current provider’s model identity.

Example

use llm::alloyed::AlloyedModelProvider;
use llm::StreamingModelProvider;

fn create_alloyed(providers: Vec<Box<dyn StreamingModelProvider>>) -> AlloyedModelProvider {
    AlloyedModelProvider::new(providers)
    // First stream_response() uses providers[0],
    // second uses providers[1], etc.
}

Created automatically by ModelProviderParser::parse when given a comma-separated model string like "anthropic:claude-sonnet-4-5-20250929,openai:gpt-4o".

Implementations

impl AlloyedModelProvider

pub fn new(providers: Vec<Box<dyn StreamingModelProvider>>) -> Self

Trait Implementations

impl StreamingModelProvider for AlloyedModelProvider

fn stream_response(&self, context: &Context) -> LlmResponseStream

fn model(&self) -> Option<LlmModel>

The LlmModel this provider is currently configured to use. Returns None for providers where the model is unknown at compile time (e.g. test fakes).

fn display_name(&self) -> String

fn context_window(&self) -> Option<u32>

Context window size in tokens for the current model. Returns None for unknown models (e.g. Ollama, LlamaCpp).