---
title: "Late Chunk"
method: POST
path: "/v1/chunk/late"
tags: ["Chunking"]
---

# Late Chunk

`POST /v1/chunk/late`

Late chunking endpoint using Modal.

Chunks text using late chunking approach with embeddings computed at chunk boundaries.
This endpoint calls the Modal function for GPU-accelerated processing.

## Headers

- `authorization` string, nullable

## Request body

- LateChunkerRequest — Request model for late chunking.
  - `text` union — Text or list of texts to chunk
    - string
    - string[]
  - `file` FileReference — File reference for chunking with OCR.
    - `type` string, required — File type: 'document' or 'base64;extension'
    - `content` string, required — Filename (for document) or base64 data
  - `ocr` string — OCR provider (mistral or gemini)
  - `embedding_model` string — Embedding model
  - `chunk_size` integer — Size of each chunk
  - `recipe` string — Chunking recipe to use
  - `lang` string — Language code
  - `min_characters_per_chunk` integer — Minimum characters per chunk

## Response `200`

Successful Response

- unknown

## Other responses

- `422` — Validation Error

---

[API](https://skmtc.dev/chonkie/apis/chonkie-api.md) · [All operations](https://skmtc.dev/chonkie/apis/chonkie-api/llms.txt) · [OpenAPI document](https://skmtc.dev/chonkie/apis/chonkie-api/revisions/f40f036f941c?raw)