Struct OllamaClient 

pub struct OllamaClient { /* private fields */ }

Client for Ollama local LLM inference

API Details

• Base URL: http://localhost:11434/api (default)
• No rate limit (local service)
• No API key required
• Falls back to mock data when Ollama is not running

Implementations

impl OllamaClient

pub fn new() -> Self

Create a new Ollama client with default base URL

pub fn with_base_url(base_url: &str) -> Self

Create a new Ollama client with custom base URL

Arguments

• base_url - Ollama API base URL (e.g., “http://localhost:11434/api”)

pub async fn is_available(&self) -> bool

Check if Ollama is available

pub async fn list_models(&mut self) -> Result<Vec<OllamaModel>>

List available models

pub async fn generate(&mut self, model: &str, prompt: &str) -> Result<String>

Generate text completion

Arguments

• model - Model name (e.g., “llama2”, “mistral”)
• prompt - Prompt text

pub async fn chat( &mut self, model: &str, messages: Vec<OllamaChatMessage>, ) -> Result<String>

Chat completion with message history

Arguments

• model - Model name
• messages - Chat message history

pub async fn embeddings( &mut self, model: &str, prompt: &str, ) -> Result<Vec<f32>>

Generate embeddings for text

Arguments

• model - Model name (e.g., “llama2”)
• prompt - Text to embed

pub async fn pull_model(&mut self, name: &str) -> Result<bool>

Pull a model from Ollama library

Arguments

• name - Model name to pull

Note

This is a blocking operation that may take several minutes

pub fn model_to_vector(&self, model: &OllamaModel) -> SemanticVector

Convert Ollama model to SemanticVector

Trait Implementations

impl Default for OllamaClient

fn default() -> Self

Returns the “default value” for a type.