Text-to-image generation

Using a coding agent? Install the together-ai skill to let your agent write correct image generation code automatically. See agent skills for details.

Generate an image

To query an image model, use the .images method and specify the image model:

TypeScript examples that call together.images.generate require [email protected] or later. If your project already has an older SDK installed, upgrade with npm install together-ai@latest.

from together import Together

client = Together()

# Generate an image from a text prompt
response = client.images.generate(
    prompt="A serene mountain landscape at sunset with a lake reflection",
    model="black-forest-labs/FLUX.2-dev",
    steps=20,
)

print(f"Image URL: {response.data[0].url}")

import Together from "together-ai";

const together = new Together();

async function main() {
  const response = await together.images.generate({
    prompt: "A serene mountain landscape at sunset with a lake reflection",
    model: "black-forest-labs/FLUX.2-dev",
    steps: 20,
  });

  console.log(response.data[0].url);
}

main();

curl -X POST "https://api.together.ai/v1/images/generations" \
     -H "Authorization: Bearer $TOGETHER_API_KEY" \
     -H "Content-Type: application/json" \
     -d '{
       "model": "black-forest-labs/FLUX.2-dev",
       "prompt": "A serene mountain landscape at sunset with a lake reflection",
       "steps": 20
     }'

Example response structure and output:

{
  "id": "oFuwv7Y-2kFHot-99170ebf9e84e0ce-SJC",
  "model": "black-forest-labs/FLUX.2-dev",
  "data": [
    {
      "index": 0,
      "url": "https://api.together.ai/v1/images/..."
    }
  ]
}

Generated image of a serene mountain landscape at sunset with a lake reflection.

Save to disk

The recommended pattern for saving a generated image locally is response_format="base64". This avoids a separate download step and works without any special HTTP headers.

import base64

from together import Together

client = Together()

response = client.images.generate(
    prompt="A serene mountain landscape at sunset with a lake reflection",
    model="black-forest-labs/FLUX.2-dev",
    steps=20,
    response_format="base64",
)

image_data = base64.b64decode(response.data[0].b64_json)
with open("output.png", "wb") as f:
    f.write(image_data)
print(f"Saved {len(image_data)} bytes to output.png")

import { writeFileSync } from "fs";
import Together from "together-ai";

const together = new Together();

const response = await together.images.generate({
  prompt: "A serene mountain landscape at sunset with a lake reflection",
  model: "black-forest-labs/FLUX.2-dev",
  steps: 20,
  response_format: "base64",
});

const imageData = Buffer.from(response.data[0].b64_json!, "base64");
writeFileSync("output.png", imageData);
console.log(`Saved ${imageData.length} bytes to output.png`);

If you use the returned url instead, download it with a non-empty User-Agent header. The CDN rejects requests with a blank User-Agent (HTTP 403):

Python

import urllib.request

req = urllib.request.Request(
    response.data[0].url,
    headers={"User-Agent": "my-app/1.0"},
)
with urllib.request.urlopen(req) as r:
    with open("output.png", "wb") as f:
        f.write(r.read())

Supported models

For the current list of image-generation models, see the serverless catalog or the dedicated model inference catalog. The table below shows the most commonly used serverless image models.

Model	API string	Serverless	Notes
FLUX.1.1 Pro	`black-forest-labs/FLUX.1.1-pro`	✓	High quality; good starting point.
FLUX.2 Pro	`black-forest-labs/FLUX.2-pro`	✓	High fidelity; supports reference images.
FLUX.2 Dev	`black-forest-labs/FLUX.2-dev`	✓	Supports `guidance`, `steps`, and low-rank adapters (LoRAs).
FLUX.2 Flex	`black-forest-labs/FLUX.2-flex`	✓	Adjustable steps/guidance; best typography.
FLUX.1 Kontext Pro	`black-forest-labs/FLUX.1-kontext-pro`	✓	Image editing via `image_url`.
FLUX.2 Max	`black-forest-labs/FLUX.2-max`	✓	Highest fidelity FLUX.2 variant.
FLUX.1 Kontext Max	`black-forest-labs/FLUX.1-kontext-max`	✓	Highest-quality image editing.

Parameters

The image generation endpoint accepts a common set of parameters across models, including prompt, model, width, height, n, steps, seed, and negative_prompt. The full parameter reference also covers image_url, reference_images, frame_images, and model-specific notes. A few quick notes:

prompt is required for all models except Kling.
width and height rely on defaults unless otherwise specified. Available dimensions differ by model.
FLUX Schnell and Kontext (Pro/Max/Dev) models use the aspect_ratio parameter to set the output image size, whereas FLUX.1 Pro, FLUX 1.1 Pro, and FLUX.1 Dev use width and height.

Generate multiple variations

Generate multiple variations of the same prompt and choose between them:

response = client.images.generate(
    prompt="A cute robot assistant helping in a modern office",
    model="black-forest-labs/FLUX.2-dev",
    n=4,
    steps=20,
)

print(f"Generated {len(response.data)} variations")
for i, image in enumerate(response.data):
    print(f"Variation {i+1}: {image.url}")

const response = await together.images.generate({
  prompt: "A cute robot assistant helping in a modern office",
  model: "black-forest-labs/FLUX.2-dev",
  n: 4,
  steps: 20,
});

console.log(`Generated ${response.data.length} variations`);

response.data.forEach((image, i) => {
  console.log(`Variation ${i + 1}: ${image.url}`);
});

curl -X POST "https://api.together.ai/v1/images/generations" \
     -H "Authorization: Bearer $TOGETHER_API_KEY" \
     -H "Content-Type: application/json" \
     -d '{
       "model": "black-forest-labs/FLUX.2-dev",
       "prompt": "A cute robot assistant helping in a modern office",
       "n": 4,
       "steps": 20
     }'

Example output:

Next steps

Image-to-image generation

Edit or transform an existing image with image_url or reference_images.

Image generation parameters

Look up the full parameter reference, custom dimensions, quality control, base64 responses, the safety checker, and troubleshooting.

GET STARTED

SERVERLESS

INFERENCE APIS

DEDICATED MODEL INFERENCE

DEDICATED CONTAINER INFERENCE

POST-TRAINING

FINE-TUNING

GPU CLUSTERS

CODE EXECUTION

ADMINISTRATION

Generate an image

Save to disk

Supported models

Parameters

Generate multiple variations

Next steps

Image-to-image generation

Image generation parameters

​Generate an image

​Save to disk

​Supported models

​Parameters

​Generate multiple variations

​Next steps

Image-to-image generation

Image generation parameters

Generate an image

Save to disk

Supported models

Parameters

Generate multiple variations

Next steps