Struct UnifiedLlmClient 

pub struct UnifiedLlmClient { /* private fields */ }

Unified LLM client supporting 18+ providers

Implementations

impl UnifiedLlmClient

pub fn new(config: LlmConfig) -> Result<Self>

Create a new client with configuration

Uses pooled HTTP clients to maintain connection reuse across calls. This eliminates TLS handshake overhead (100-500ms per call).

pub fn default_anthropic() -> Result<Self>

Create with default configuration (Anthropic Claude)

pub fn openai(model: impl Into<String>) -> Result<Self>

Create for OpenAI

pub fn openrouter(model: impl Into<String>) -> Result<Self>

Create for OpenRouter (300+ models)

pub fn gemini(model: impl Into<String>) -> Result<Self>

Create for Google Gemini (AI Studio)

pub fn grok(model: impl Into<String>) -> Result<Self>

Create for xAI Grok

pub fn groq(model: impl Into<String>) -> Result<Self>

Create for Groq (ultra-fast)

pub fn mistral(model: impl Into<String>) -> Result<Self>

Create for Mistral

pub fn deepseek(model: impl Into<String>) -> Result<Self>

Create for DeepSeek

pub fn together(model: impl Into<String>) -> Result<Self>

Create for Together AI

pub fn fireworks(model: impl Into<String>) -> Result<Self>

Create for Fireworks AI

pub fn qwen(model: impl Into<String>) -> Result<Self>

Create for Alibaba Qwen

pub fn cohere(model: impl Into<String>) -> Result<Self>

Create for Cohere

pub fn perplexity(model: impl Into<String>) -> Result<Self>

Create for Perplexity

pub fn cerebras(model: impl Into<String>) -> Result<Self>

Create for Cerebras

pub fn azure( resource: impl Into<String>, deployment: impl Into<String>, model: impl Into<String>, ) -> Result<Self>

Create for Azure OpenAI

pub fn cloudflare_gateway( account_id: impl Into<String>, gateway_id: impl Into<String>, model: impl Into<String>, ) -> Result<Self>

Create for Cloudflare AI Gateway

pub fn ollama() -> Result<Self>

Create for Ollama (local server; can route to Ollama Cloud models)

Uses env var resolution (preferred):

• RK_OLLAMA_MODEL (fallback: OLLAMA_MODEL)
• RK_OLLAMA_URL (fallback: OLLAMA_URL, default: http://localhost:11434)

Trait Implementations

impl LlmClient for UnifiedLlmClient

fn complete<'life0, 'async_trait>( &'life0 self, request: LlmRequest, ) -> Pin<Box<dyn Future<Output = Result<LlmResponse>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait,

Generate a completion

fn provider(&self) -> LlmProvider

Get the provider

fn model(&self) -> &str

Get the model name