Semantic Search with Cloudflare Workers AI Embeddings

Set up Meilisearch with Cloudflare Workers AI embedding models for semantic search.

Cloudflare Workers AI provides embedding models that run on Cloudflare’s edge network. This guide shows you how to configure Meilisearch with Cloudflare Workers AI embeddings using the REST embedder.

Requirements

• A Meilisearch project
• A Cloudflare account with access to Workers AI
• Your Cloudflare account ID and API key

Available models

Model	Dimensions	Notes
@cf/baai/bge-small-en-v1.5	384	Fastest, English only
@cf/baai/bge-base-en-v1.5	768	Balanced, English only
@cf/baai/bge-large-en-v1.5	1024	Highest quality BGE, English only
@cf/google/embeddinggemma-300m	768	Google’s compact embedding model
@cf/qwen/qwen3-embedding-0.6b	1024	Qwen3’s lightweight embedding model

Configure the embedder

Update your index settings with the Cloudflare Workers AI embedder configuration:

{
  "cloudflare": {
    "source": "rest",
    "apiKey": "<CLOUDFLARE_API_KEY>",
    "dimensions": 384,
    "documentTemplate": "A product named '{{doc.name}}': {{doc.description}}",
    "url": "https://api.cloudflare.com/client/v4/accounts/<ACCOUNT_ID>/ai/run/@cf/baai/bge-small-en-v1.5",
    "request": {
      "text": ["{{text}}", "{{..}}"]
    },
    "response": {
      "result": {
        "data": ["{{embedding}}", "{{..}}"]
      }
    }
  }
}

Replace <CLOUDFLARE_API_KEY> with your Cloudflare API key and <ACCOUNT_ID> with your Cloudflare account ID. The model name is part of the URL path. Adjust dimensions to match the model you choose.

Send this configuration to Meilisearch:

curl \
  -X PATCH 'MEILISEARCH_URL/indexes/INDEX_NAME/settings' \
  -H 'Content-Type: application/json' \
  -H 'Authorization: Bearer MEILISEARCH_KEY' \
  --data-binary '{
    "embedders": {
      "cloudflare": {
        "source": "rest",
        "apiKey": "<CLOUDFLARE_API_KEY>",
        "dimensions": 384,
        "documentTemplate": "A product named '\''{{doc.name}}'\'': {{doc.description}}",
        "url": "https://api.cloudflare.com/client/v4/accounts/<ACCOUNT_ID>/ai/run/@cf/baai/bge-small-en-v1.5",
        "request": {
          "text": ["{{text}}", "{{..}}"]
        },
        "response": {
          "result": {
            "data": ["{{embedding}}", "{{..}}"]
          }
        }
      }
    }
  }'

Meilisearch handles batching and rate limiting automatically. Monitor the tasks queue to track indexing progress.

Test the search

{
  "q": "comfortable shoes for walking",
  "hybrid": {
    "semanticRatio": 0.5,
    "embedder": "cloudflare"
  }
}

Next steps

• Document template best practices to optimize which fields are embedded
• Custom hybrid ranking to tune the balance between keyword and semantic results
• Embedder settings reference for all configuration options