embeddinggemma-2

export interface Env {
	AI: Ai;
}

export default {
	async fetch(request, env): Promise<Response> {
		// Prefix queries and documents with task instructions for better retrieval quality.
		const response = await env.AI.run("@cf/google/embeddinggemma-2", {
			input: [
				"task: search result | query: waterproof shoes for trail running",
				"title: Trail Runner X | text: Waterproof running shoes with a breathable mesh upper.",
				"title: City Sneaker | text: Lightweight canvas sneakers for everyday wear.",
			],
		});

		// response.data[i].embedding is a 768-dimensional vector
		return Response.json(response);
	},
} satisfies ExportedHandler<Env>;

import os
import requests

ACCOUNT_ID = "your-account-id"
AUTH_TOKEN = os.environ.get("CLOUDFLARE_AUTH_TOKEN")

response = requests.post(
    f"https://api.cloudflare.com/client/v4/accounts/{ACCOUNT_ID}/ai/run/@cf/google/embeddinggemma-2",
    headers={"Authorization": f"Bearer {AUTH_TOKEN}"},
    json={
        "input": [
            "task: search result | query: waterproof shoes for trail running",
            "title: Trail Runner X | text: Waterproof running shoes with a breathable mesh upper.",
            "title: City Sneaker | text: Lightweight canvas sneakers for everyday wear.",
        ]
    },
)
print(response.json())

curl https://api.cloudflare.com/client/v4/accounts/$CLOUDFLARE_ACCOUNT_ID/ai/run/@cf/google/embeddinggemma-2 \
  -X POST \
  -H "Authorization: Bearer $CLOUDFLARE_AUTH_TOKEN" \
  -d '{
    "input": [
      "task: search result | query: waterproof shoes for trail running",
      "title: Trail Runner X | text: Waterproof running shoes with a breathable mesh upper.",
      "title: City Sneaker | text: Lightweight canvas sneakers for everyday wear."
    ]
  }'

export interface Env {
	AI: Ai;
}

// Media must be sent as base64 data URIs.
async function toDataUri(url: string, mimeType: string): Promise<string> {
	const bytes = new Uint8Array(await (await fetch(url)).arrayBuffer());
	let binary = "";
	for (const byte of bytes) binary += String.fromCharCode(byte);
	return `data:${mimeType};base64,${btoa(binary)}`;
}

export default {
	async fetch(request, env): Promise<Response> {
		const image = await toDataUri("https://example.com/shoe.png", "image/png");
		const video = await toDataUri("https://example.com/grip-test.mp4", "video/mp4");

		// Each conversation in the array produces one embedding.
		const response = await env.AI.run("@cf/google/embeddinggemma-2", {
			messages: [
				// Text, an image, and a video combined into a single embedding
				[
					{
						role: "user",
						content: [
							{ type: "text", text: "Waterproof running shoes." },
							{ type: "image_url", image_url: { url: image } },
							{ type: "video_url", video_url: { url: video } },
						],
					},
				],
				// A text query to compare against
				[
					{
						role: "user",
						content: "task: search result | query: waterproof shoes for trail running",
					},
				],
			],
		});

		// Both embeddings share one vector space, so you can compare them directly.
		return Response.json(response);
	},
} satisfies ExportedHandler<Env>;

import base64
import os
import requests

ACCOUNT_ID = "your-account-id"
AUTH_TOKEN = os.environ.get("CLOUDFLARE_AUTH_TOKEN")

def data_uri(path, mime_type):
    with open(path, "rb") as f:
        return f"data:{mime_type};base64," + base64.b64encode(f.read()).decode()

response = requests.post(
    f"https://api.cloudflare.com/client/v4/accounts/{ACCOUNT_ID}/ai/run/@cf/google/embeddinggemma-2",
    headers={"Authorization": f"Bearer {AUTH_TOKEN}"},
    json={
        "messages": [
            [
                {
                    "role": "user",
                    "content": [
                        {"type": "text", "text": "Waterproof running shoes."},
                        {"type": "image_url", "image_url": {"url": data_uri("shoe.png", "image/png")}},
                        {"type": "video_url", "video_url": {"url": data_uri("grip-test.mp4", "video/mp4")}},
                    ],
                }
            ],
            [
                {
                    "role": "user",
                    "content": "task: search result | query: waterproof shoes for trail running",
                }
            ],
        ]
    },
)
print(response.json())

export interface Env {
	AI: Ai;
}

export default {
	async fetch(request, env): Promise<Response> {
		// Use mono 16 kHz audio for best results.
		const bytes = new Uint8Array(await (await fetch("https://example.com/clip.wav")).arrayBuffer());
		let binary = "";
		for (const byte of bytes) binary += String.fromCharCode(byte);

		const response = await env.AI.run("@cf/google/embeddinggemma-2", {
			messages: [
				{
					role: "user",
					content: [
						{ type: "input_audio", input_audio: { data: btoa(binary), format: "wav" } },
					],
				},
			],
			dimensions: 256,
		});

		// dimensions: 256 returns a shorter vector to reduce storage.
		return Response.json(response);
	},
} satisfies ExportedHandler<Env>;

import base64
import os
import requests

ACCOUNT_ID = "your-account-id"
AUTH_TOKEN = os.environ.get("CLOUDFLARE_AUTH_TOKEN")

with open("clip.wav", "rb") as f:
    audio = base64.b64encode(f.read()).decode()

response = requests.post(
    f"https://api.cloudflare.com/client/v4/accounts/{ACCOUNT_ID}/ai/run/@cf/google/embeddinggemma-2",
    headers={"Authorization": f"Bearer {AUTH_TOKEN}"},
    json={
        "messages": [
            {
                "role": "user",
                "content": [
                    {"type": "input_audio", "input_audio": {"data": audio, "format": "wav"}}
                ],
            }
        ],
        "dimensions": 256,
    },
)
print(response.json())