Document, media and web processing for companies. Prices are quoted in USD and paid in native USDC. Every operation is billed per measured unit from verified unit costs and never consumes more than the quoted maximum.
| Operation | Price | Cost | Status |
|---|---|---|---|
| Loading live pricing… | |||
Live values above come from /agent-tools/api/pricing.
| Operation | What it does |
|---|---|
| web_fetch | Fetch a public HTTPS page and return its visible text. |
| web_extract | Extract structured text from a public HTTPS page. |
| web_monitor | Detect whether a page changed since a previous content hash. |
| pdf_to_text | Extract text from PDF documents. |
| pdf_split | Split a PDF into a selected page range. |
| pdf_merge | Merge up to five PDF inputs into one document. |
| document_ocr | OCR scanned documents into text. |
| audio_transcribe | Transcribe audio into text. |
The three web operations still accept a public HTTPS url exactly as before, and now also accept a mutually exclusive Google query. Search is answered by Serper and returns only real https source URLs. An optional "diagnose": true adds a separately priced DeepSeek interpretation with cited evidence. Diagnostics never replace the primary result: when the model is unavailable the job still completes and reports diagnostic_status: unavailable. Measured provider spend is recorded per job against a daily per-company budget.
1 · Create a company workspace at the dashboard — it starts with 1,000 trial credits.
2 · Issue a scoped API key from the dashboard. The secret is shown once and stored only as a hash.
3 · Upload a document, request a quote, submit a job and track it.
4 · Download the result from the dashboard or the results API.
Paid credits are bought with native USDC on Polygon PoS (Base is also supported) and never expire.