Question 1

What are tokens and why do they matter?

Accepted Answer

Tokens are the fundamental units AI models use to process text. Most AI APIs charge per token, so knowing your count helps estimate costs and stay within context limits.

Question 2

How accurate is the token count?

Accepted Answer

For OpenAI models, exact via tiktoken WebAssembly. For others, smart approximation marked with ~.

Question 3

Is my data safe?

Accepted Answer

Your prompts never leave your browser. Tokenization runs locally via WebAssembly. Only anonymous page views are tracked with consent.

Question 4

How does the prompt optimizer work?

Accepted Answer

Detects verbose phrases, filler words, and repetitions. Auto-optimize applies safe changes with one click and undo.

Question 5

What AI models are supported?

Accepted Answer

16 models: GPT-4o, GPT-4o mini, GPT-4 Turbo, GPT-3.5, o1, o1-mini, Claude 3.5 Sonnet, Claude 3.5 Haiku, Claude 3 Opus, Gemini 1.5 Pro, Gemini 1.5 Flash, Gemini 2.0 Flash, Mistral Large, Mistral Small, Llama 3.1 70B.

Question 6

How does the Ratio Calculator work?

Accepted Answer

Paste a real prompt and response to measure your output ratio. Get specific advice to reduce costs: length constraints, structured output, no-preamble instructions.

Question 7

What are prompt variables?

Accepted Answer

Write {{name}} in your prompt. A panel appears to fill values. Token count reflects the resolved text. The template stays reusable.

Question 8

Can I use this offline?

Accepted Answer

Basic functionality works offline after first visit. The tiktoken WASM module is cached by your browser.