Practical tools for building with LLMs
MakeTools is a growing set of free tools for people who build with large language models: estimate token costs, check whether a prompt fits a context window, size a GPU or a vector index, split documents for retrieval, and convert tool definitions between providers. Everything runs in your browser, so your prompts and data are never uploaded.
Who it is for
Developers and engineers who need a quick, checkable answer before they commit time or money: “what will this cost per month?”, “will it fit on my GPU?”, “how big will this index be?”. Each tool shows its formula and its limits, and every price or model limit it uses is linked to the provider’s own page with the date it was checked.
LLM engineering
Calculators and converters for building with LLMs: token cost, context windows, caching, VRAM, vector size, chunking, tool schemas, JSONL data and rate limits.
LLM Token Cost Calculator
Tokens in, cost per call, per 1,000 calls and per month.
Context Window Checker
Will your prompt, chunks and answer fit the context window?
Prompt Caching Calculator
Cached vs uncached input cost, and the break-even point.
Local LLM VRAM Calculator
Weights + KV cache + overhead, and whether it fits your GPU.
Vector DB Size Calculator
RAM and disk for N vectors at a given precision.
RAG Text Chunker
Split text by characters, tokens or sentences with overlap.
Tool Schema Converter
OpenAI ⇄ Anthropic ⇄ Gemini ⇄ MCP tool definitions.
Fine-tuning JSONL Validator
Line-by-line checks for chat-format training data.
LLM Rate Limit Planner
Max requests per minute, concurrency and queue time.
Data & code conversion
Convert a T-SQL CREATE TABLE statement to PostgreSQL DDL, or turn a JSON sample into Pydantic v2 models. Deterministic, documented rules, no AI and no uploads.
Newest guides
Choosing chunk size and overlap for RAG
How chunk size and overlap affect retrieval quality, cost and context use, with chunk-count arithmetic and a method for testing settings on your own questions.
Function-calling formats compared
The same tool definition written in OpenAI, Anthropic, Gemini and MCP formats, what each field is called, and what gets lost when you convert between them.
How LLM API pricing works: input, output and cached tokens
What you are billed for when calling an LLM API: input, output, cache write and cache read tokens, the price modifiers, and a worked monthly cost example.
How the site works
- Local processing. Your text, files and numbers stay in your browser tab.
- No sign-up, no tracking. There are no accounts, no analytics and no cookies.
- Sourced data. Prices and model limits carry a source link and a “checked on” date. Anything that could not be verified is left out.
- Works offline. After your first visit the tools keep working without a connection.
Read about MakeTools, the editorial policy, or the changelog of tool additions and data updates.