Struct TokenCounterFactory 

pub struct TokenCounterFactory;

Factory for creating token counters for different providers.

Use this factory to get the appropriate token counter for your provider and model. The factory handles selecting the correct tokenizer and context window size.

Example

use multi_llm::{TokenCounterFactory, TokenCounter};

// Create counter for OpenAI
let openai = TokenCounterFactory::create_counter("openai", "gpt-4")?;

// Create counter for Anthropic
let anthropic = TokenCounterFactory::create_counter("anthropic", "claude-3-5-sonnet")?;

// Create counter with custom limit
let custom = TokenCounterFactory::create_counter_with_limit("openai", "gpt-4", 4096)?;

Supported Providers

• openai: Uses tiktoken for exact counts
• anthropic: Uses approximation with safety margin
• ollama: Uses cl100k_base (approximation)
• lmstudio: Uses cl100k_base (approximation)

Implementations

impl TokenCounterFactory

pub fn create_counter( provider: &str, model: &str, ) -> LlmResult<Arc<dyn TokenCounter>>

Create token counter for specific provider and model

pub fn create_counter_with_limit( provider: &str, model: &str, max_tokens: u32, ) -> LlmResult<Arc<dyn TokenCounter>>

Create counter with custom context window size