AI and LLM Tools

Token Counter, LLM Cost Calculator, Context Window and RAG Chunk Calculators

Practical tools for anyone building with or writing for large language models. Count tokens, price API calls across 82 models, check what fits in a context window, plan retrieval chunks, structure and compare prompts, and clean Markdown out of chatbot replies, all in your browser.

Every AI tool

AI Token Counter

Count GPT tokens in a prompt or document with o200k_base, in your browser.

Context Window Calculator

Check whether a prompt, documents and reply fit a model's context window.

Embedding Cost Calculator

Estimate what it costs to embed documents and how much storage the vectors take.

LLM Cost Calculator

Price LLM API calls per request and per month, with caching.

Markdown Cleaner

Strip Markdown from AI output and get clean plain text.

Prompt Diff

Compare two prompt versions and see exactly what changed.

Prompt Formatter

Turn loose notes into a structured prompt with XML tags or headings.

RAG Chunk Calculator

Plan chunk size and overlap for retrieval-augmented generation.

Tokens and context

The AI token counter counts tokens with the o200k_base encoding used by recent OpenAI models, along with characters and words; other providers tokenize differently, so treat it as a guide for them. The context window calculator checks whether a system prompt, conversation and documents fit a model's limit with room left for the reply.

Costs

The LLM cost calculator prices requests per call, per 1,000 and per month for 82 models from the model prices dataset, including cached input where providers offer it, or your own rates. The embedding cost calculator does the same for indexing a document collection.

Retrieval

The RAG chunk calculator works out how many chunks a corpus splits into for a chosen chunk size and overlap, and how many retrieved chunks fit in a prompt.

Prompts and output

The prompt formatter wraps a role, task, context, constraints and output format in XML tags or headings. Prompt diff compares two prompt versions line by line, and the Markdown cleaner strips formatting syntax from model replies to leave plain text.