Próximamente: upgrade a Pro y Team.
Know what your AI calls will cost before you make them. Compare across OpenAI, Anthropic, Google, Meta and Mistral.
Compare models
| Model | Tokens | Context used | Input cost | Output cost | Precision |
|---|
* Output cost assumes reply ≈ same length as input. Costs based on public pricing and may be outdated.
Removes layout noise from PDFs, DOCX and HTML without touching meaning. Typical savings: 10–40%.
Original chars
—
Cleaned chars
—
Chars removed
—
Savings
—
Token savings per model:
| Model | Before | After | Saved | % |
|---|
Cleaned text (ready to paste):
Upload a large doc + your question — returns only relevant chunks. TF-IDF keyword search. Best for focused queries.
Token savings per model:
| Model | Full doc | Retrieved | Saved | % |
|---|
Relevant passages (sorted by relevance):
Tokens are the chunks LLMs process — roughly 3–4 characters of English text. Code and non-Latin scripts often produce more tokens per character.
OpenAI uses the open-source tiktoken library — we run the same tokenizer server-side, so counts are exact. Anthropic, Google and others use proprietary tokenizers; we estimate and label it clearly.
Cost = (tokens ÷ 1,000,000) × price-per-million-tokens. Input and output tokens are billed separately by most providers.
Everything you need to ship AI features without surprises.
Free
$0/mo
Pro
$12/mo