---
title: RAG Chunking API Pricing ⬣ POMA AI
description: PrimeCut adaptive pricing, max €0.01/page — documents, audio, video. Free tier with 1,000 pages on sign-up, and Enterprise with dedicated VPC and a signed DPA.
canonical: https://www.poma-ai.com/pricing
generated: Markdown variant of the page above, built from the prerendered HTML
---
Drop PrimeCut into your existing RAG pipeline and pay only for what you process.

Every account starts with 1,000 free pages on sign-up — no credit card required to try PrimeCut.

Free

### Trial

Try PrimeCut on real documents at no cost. Drop the SDK in, point it at a document, see what comes back. No credit card required.

Free

Try for free

1,000 free pages. Up to 250 pages per document.

Pay-as-you-go

### On-demand

One API that adapts processing to each document — preserving hierarchy, cross-references, and visual content where it matters, and staying lean where it doesn't.

Adaptive max €0.01 / page / 1000 token / min

Get started

Adaptive processing on every API request — one balance, priced per page. A page is one document page, one minute of audio or video, or 1,000 tokens of delivered content, whichever counts higher.

Custom

### Enterprise

For teams of any size that need privacy, control, and security. Deploy POMA in a dedicated instance, your own VPC, or on-premise, including isolated networks, with multi-user access, full data isolation, dedicated technical support, and pricing tailored to your needs.

Custom

Contact sales

Multi-user, dedicated VPC or on-premise, signed custom DPA.

Looking for Grill?

## Grill is now index4.ai

The managed context engine has its own site and its own plans — ingestion, chunking, retrieval and ranking, run for you on the same PrimeCut core. [See index4.ai pricing (opens in a new tab)](https://index4.ai/pricing)

What counts as a page?

-   **Documents** — one page, or 1,000 tokens of text for a dense page — whichever is higher.
-   **Audio & video** — one minute, or 1,000 tokens — whichever is higher.
-   **Pageless files** (text, HTML, spreadsheets, JSON, …) — every 1,000 tokens of extracted content.

We count only the text you actually receive — your finished chunks, tables, and image descriptions — so easy documents cost less. The token count only matters for very dense pages or video minutes (more than 1,000 tokens of output).

## Features

What PrimeCut includes.

### Ingestion & parsing

-   OCR-aware ingestion preserves tables and visual content.
-   Hierarchical chunking with structural fidelity.
-   50+ filetypes across documents, presentations, spreadsheets, images, and more.

Documents

`pdf``doc``docx``dotx``rtf``txt``md``html``htm``xml`

Presentations

`ppt``pptx``pps``ppsx``pot``potx``key`

Spreadsheets

`xls``xlsx``xlsb``xltx``csv``numbers``ods``odc`

Images

`png``jpg``jpeg``gif``bmp``tif``tiff``svg``webp``ico``heic``heif``psd`

Audio & video

`mp4``mov``webm``mkv``avi``m4v``mpeg``mpg``mp3``wav``m4a``aac``ogg``flac``opus`

Other

`epub``mobi``djvu``dwg``dxf``dwf``dwfx``vsd``vsdx``ai``eps``ps``prn``xps``oxps``pub``mdi``pages``odp``odf``odt`

### Chunking & output

-   Adaptive processing on every request — matched to each document's complexity.
-   Structured files (`xml`, `cir`, `json`, `yaml`, `toml`, `ini`, `env`, `csv`, `tsv`, `xls`, `xlsx`, `xlsb`) are always chunked with full structural fidelity.
-   77% fewer tokens at 100% recall on benchmarked datasets.
-   Chunks returned with hierarchy intact for downstream embedding.

### Integration

-   LangChain and LlamaIndex compatible.
-   REST API + SDKs; drop into existing RAG pipelines.

### Enterprise extras

-   Dedicated VPC, custom DPA, multi-user controls.
-   On-premise deployment: run PrimeCut in your own datacenter or an isolated network, scoped with us.
-   Priority support and onboarding on Enterprise.

### Supported languages

-   Ingest and semantic search work across every supported language. Full-text search quality varies by language — see the table for the tier each one falls into.

-   Stemmed — best quality, language-aware matching
-   Exact-match — literal token matching
-   Char-bigram — for scripts without word boundaries

Language

Code

Ingest

Semantic search

Full-text search

Afrikaans

`af`

✅

✅

exact-match

Albanian

`sq`

✅

✅

exact-match

Amharic

`am`

✅

✅

exact-match

Arabic

`ar`

✅

✅

stemmed

Armenian

`hy`

✅

✅

stemmed

Azerbaijani

`az`

✅

✅

exact-match

Basque

`eu`

✅

✅

stemmed

Bavarian

`bar`

✅

✅

stemmed (via German)

Belarusian

`be`

✅

✅

exact-match

Bengali

`bn`

✅

✅

exact-match

Bulgarian

`bg`

✅

✅

exact-match

Burmese

`my`

✅

✅

char-bigram

Catalan

`ca`

✅

✅

stemmed

Chinese

`zh`

✅

✅

char-bigram

Croatian

`hr`

✅

✅

exact-match

Czech

`cs`

✅

✅

exact-match

Danish

`da`

✅

✅

stemmed

Dutch

`nl`

✅

✅

stemmed

English

`en`

✅

✅

stemmed

Estonian

`et`

✅

✅

stemmed

Finnish

`fi`

✅

✅

stemmed

French

`fr`

✅

✅

stemmed

Galician

`gl`

✅

✅

exact-match

Georgian

`ka`

✅

✅

exact-match

German

`de`

✅

✅

stemmed

Greek

`el`

✅

✅

stemmed

Gujarati

`gu`

✅

✅

exact-match

Hebrew

`he`

✅

✅

exact-match

Hindi

`hi`

✅

✅

stemmed

Hungarian

`hu`

✅

✅

stemmed

Icelandic

`is`

✅

✅

exact-match

Indonesian

`id`

✅

✅

stemmed

Irish

`ga`

✅

✅

stemmed

Italian

`it`

✅

✅

stemmed

Japanese

`ja`

✅

✅

char-bigram

Kannada

`kn`

✅

✅

exact-match

Kazakh

`kk`

✅

✅

exact-match

Khmer

`km`

✅

✅

char-bigram

Korean

`ko`

✅

✅

char-bigram

Kyrgyz

`ky`

✅

✅

exact-match

Lao

`lo`

✅

✅

char-bigram

Latvian

`lv`

✅

✅

exact-match

Lithuanian

`lt`

✅

✅

stemmed

Plattdeutsch

`nds`

✅

✅

stemmed (via German)

Malay

`ms`

✅

✅

exact-match

Malayalam

`ml`

✅

✅

exact-match

Maltese

`mt`

✅

✅

exact-match

Marathi

`mr`

✅

✅

exact-match

Mongolian

`mn`

✅

✅

exact-match

Nepali

`ne`

✅

✅

exact-match

Norwegian

`no`

✅

✅

stemmed

Pashto

`ps`

✅

✅

exact-match

Persian

`fa`

✅

✅

exact-match

Polish

`pl`

✅

✅

exact-match

Portuguese

`pt`

✅

✅

stemmed

Punjabi

`pa`

✅

✅

exact-match

Romanian

`ro`

✅

✅

stemmed

Russian

`ru`

✅

✅

stemmed

Serbian

`sr`

✅

✅

stemmed

Sindhi

`sd`

✅

✅

exact-match

Sinhala

`si`

✅

✅

exact-match

Slovak

`sk`

✅

✅

exact-match

Slovenian

`sl`

✅

✅

exact-match

Spanish

`es`

✅

✅

stemmed

Swahili

`sw`

✅

✅

exact-match

Swedish

`sv`

✅

✅

stemmed

Swiss German

`gsw`

✅

✅

stemmed (via German)

Tagalog

`tl`

✅

✅

exact-match

Tajik

`tg`

✅

✅

exact-match

Tamil

`ta`

✅

✅

stemmed

Telugu

`te`

✅

✅

exact-match

Thai

`th`

✅

✅

char-bigram

Tibetan

`bo`

✅

✅

char-bigram

Turkish

`tr`

✅

✅

stemmed

Turkmen

`tk`

✅

✅

exact-match

Ukrainian

`uk`

✅

✅

exact-match

Urdu

`ur`

✅

✅

exact-match

Uzbek

`uz`

✅

✅

exact-match

Vietnamese

`vi`

✅

✅

exact-match

Welsh

`cy`

✅

✅

exact-match

Yiddish

`yi`

✅

✅

exact-match

Yoruba

`yo`

✅

✅

exact-match

Zulu

`zu`

✅

✅

exact-match