Implementation:Cohere ai Cohere python RerankResponse Model

Field	Value
Type	Implementation
Source	Cohere Python SDK
Domain	NLP Information Retrieval Response Parsing
Last Updated	2026-02-15
Implements	Principle:Cohere_ai_Cohere_python_Rerank_Response_Processing

Overview

Concrete Pydantic models for rerank API responses with ranked results and relevance scores.

Description

V2RerankResponse and RerankResponse contain id, results (ordered list), and meta. RerankResponseResultsItem has index (int, position in original list), relevance_score (float, 0-1), and optional document field.

Code Reference

• src/cohere/types/rerank_response.py Lines L1-27
• src/cohere/types/rerank_response_results_item.py Lines L1-33
• src/cohere/v2/types/v2rerank_response.py Lines L1-27

Signature

class RerankResponse(UncheckedBaseModel):
    id: typing.Optional[str] = None
    results: typing.List[RerankResponseResultsItem]
    meta: typing.Optional[ApiMeta] = None

class RerankResponseResultsItem(UncheckedBaseModel):
    document: typing.Optional[RerankResponseResultsItemDocument] = None
    index: int
    relevance_score: float

Import

from cohere.types import RerankResponse (accessed via client.rerank() return)

Inputs

Parameter	Type	Required	Description
Raw API response	JSON	Yes	The raw JSON response from the rerank API

Outputs

results (List[RerankResponseResultsItem]) ordered by relevance, each with index (int) and relevance_score (float).

Example

response = client.rerank(model="rerank-v4.0-pro", query="query", documents=docs, top_n=5)
for r in response.results:
    print(f"Doc[{r.index}]: score={r.relevance_score:.4f}, text={docs[r.index][:50]}...")

Related

• Principle:Cohere_ai_Cohere_python_Rerank_Response_Processing