Estimate token count for LLM prompts. Uses a character-level approximation (~4 chars/token for English). Results are approximate — actual tokenization varies by model.
Count LLM tokens for OpenAI, Anthropic, and other models. Estimate API costs before sending requests.