Managed Inference and Agents API with Cohere Rerank 3.5

Table of Contents [expand]

Last updated January 15, 2026

Cohere Rerank 3.5 is a reranking model that scores documents based on their semantic relevance to a query. It offers enhanced reasoning, broad data compatibility, and multilingual support for over 100 languages while maintaining industry-leading accuracy.

  • Model ID: cohere-rerank-3-5
  • Region: us, eu

When to Use This Model

Cohere Rerank 3.5 is ideal for enhancing retrieval-augmented generation (RAG) systems and semantic search applications. You can use this model after retrieving results to identify the most relevant documents for your query. It offers multilingual support for over 100 languages.

Usage

Cohere Rerank 3.5 is available on our /v1/rerank API endpoint.

To provision access to the model, attach cohere-rerank-3-5 to your app $APP_NAME:

heroku ai:models:create -a $APP_NAME cohere-rerank-3-5 --as RERANK

Using config variables, you can invoke cohere-rerank-3-5 in a variety of ways:

Rate Limits

  • Maximum requests per minute: 250

Example curl Request

Get started quickly with an example request:

export RERANK_MODEL_ID=$(heroku config:get -a $APP_NAME RERANK_MODEL_ID)
export RERANK_KEY=$(heroku config:get -a $APP_NAME RERANK_KEY)
export RERANK_URL=$(heroku config:get -a $APP_NAME RERANK_URL)

curl $RERANK_URL/v1/rerank \
  -H "Authorization: Bearer $RERANK_KEY" \
  -d @- <<EOF
{
  "model": "$RERANK_MODEL_ID",
  "query": "How do I scale my application dynos?",
  "documents": [
    "Use the heroku ps:scale command to adjust the number of dynos running your application.",
    "Environment variables can be configured through the Heroku dashboard or CLI.",
    "Scaling dynos horizontally improves throughput by distributing load across multiple instances.",
    "Database connection limits should be considered when scaling to avoid exhausting connections."
  ],
  "top_n": 2
}
EOF