POST
/
v1
/
chunk
/
semantic
curl --request POST \
  --url https://api.chonkie.ai/v1/chunk/semantic \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: multipart/form-data' \
  --form embedding_model=minishlab/potion-base-8M \
  --form 'threshold=<string>' \
  --form chunk_size=512 \
  --form similarity_window=1 \
  --form min_sentences=1 \
  --form min_chunk_size=null \
  --form min_characters_per_sentence=12 \
  --form threshold_step=0.01 \
  --form 'delim=<string>' \
  --form include_delim=prev \
  --form return_type=chunks
[
  {
    "text": "<string>",
    "start_index": 123,
    "end_index": 123,
    "token_count": 123,
    "sentences": [
      {
        "text": "<string>",
        "start_index": 123,
        "end_index": 123,
        "token_count": 123,
        "embedding": [
          123
        ]
      }
    ]
  }
]

Authorizations

​
Authorization
string
header
required

Your API Key from the Chonkie Cloud dashboard

Body

multipart/form-data
​
file
file

The file to chunk.

​
embedding_model
string
default:minishlab/potion-base-8M

Model identifier or embedding model instance to use for semantic analysis.

​
threshold
default:auto

Similarity threshold for grouping sentences. Can be a float [0,1] for direct threshold, int (1,100] for percentile, or 'auto' for automatic calculation.

Allowed value: "auto"
​
chunk_size
integer
default:512

Maximum tokens per chunk.

​
similarity_window
integer
default:1

Number of preceding sentences to consider for similarity comparison.

​
min_sentences
integer
default:1

Minimum number of sentences per chunk.

​
min_chunk_size
integer

Minimum tokens per chunk (optional).

​
min_characters_per_sentence
integer
default:12

Minimum number of characters per sentence.

​
threshold_step
number
default:0.01

Step size used when automatically calculating the similarity threshold.

​
delim
default:[".","!","?","\n"]

Delimiters to split sentences on.

​
include_delim
enum<string> | null
default:prev

Include delimiters in the chunk text. If so, specify whether to include the previous or next delimiter.

Available options:
prev,
next
​
return_type
enum<string>
default:chunks

Return type for chunking. If 'chunks', returns a list of SemanticChunk objects. If 'texts', returns a list of strings.

Available options:
texts,
chunks

Response

200 - application/json
Successful Response: A list of `SemanticChunk` objects.
​
text
string

The actual text content of the chunk.

​
start_index
integer

The starting character index of the chunk within the original input text.

​
end_index
integer

The ending character index (exclusive) of the chunk within the original input text.

​
token_count
integer

The number of tokens in this specific chunk, according to the tokenizer used.

​
sentences
object[]

List of SemanticSentence objects contained within this chunk.

Represents a single sentence within a semantic chunk, including an optional embedding vector.