Name: ByteDance BAGEL API: Image Generation, Editing and Q&A
Brand: magicapi
SKU: api-magicapi-unified-multimodal-image-ai-bytedance-bagel
Availability: InStock
Author: API Market

Shashank Agarwal

# 🥯 Image Generation, Editing and Understanding API (ByteDance BAGEL)

## Overview
BAGEL is a multimodal model from ByteDance Seed. Its developer describes it as one 7B-parameter model that generates images, edits images and understands images.

Pick the job with `task`: `text-to-image` (default) creates an image from `prompt`, `image-editing` changes the input `image` as the `prompt` describes, and `image-understanding` answers the `prompt` about the `image` in text. Each request returns an object with an `image` URL and a `text` field.

## Features
- **Three tasks in one API**: `text-to-image`, `image-editing` and `image-understanding`.
- **Instruction editing**: describe the change in plain words, for example a new place or action for the same person.
- **Thinking mode**: `enable_thinking` turns on chain-of-thought reasoning before the result.
- **Guidance controls**: `cfg_text_scale` (1 to 20) for prompt adherence and `cfg_img_scale` (1 to 10) for keeping input image details.
- **Sampling controls**: `num_inference_steps` (1 to 100), `timestep_shift` (1 to 10), `cfg_renorm_type` (`global`, `local`, `text_channel`) and `cfg_renorm_min` (0 to 1).
- **Output format**: `webp` (default), `jpg` or `png`, with `output_quality` 1 to 100.

## Use Cases
1. **Character consistency**: place the same person in new scenes while keeping their clothes and look.
2. **Product and marketing images**: generate scenes from text, then edit them with follow-up instructions.
3. **Photo edits by instruction**: change the setting, action or details of a photo without masks.
4. **Image questions**: ask what an image shows and get a text answer with `image-understanding`.

## Visual Examples
### Example Input
An `image-editing` request: a photo of a woman and the instruction "She boards a modern London Tube, quietly reading a folded newspaper, wearing the same clothes", with `cfg_img_scale` 2, `cfg_renorm_type` `text_channel` and 50 steps:

![Input image: the reference photo of the woman to edit](https://blog.api.market/wp-content/uploads/2026/10/bagel-input-api-market.png)

```json
{
  "task": "image-editing",
  "image": "https://blog.api.market/wp-content/uploads/2026/10/bagel-input-api-market.png",
  "prompt": "She boards a modern London Tube, quietly reading a folded newspaper, wearing the same clothes",
  "cfg_img_scale": 2,
  "output_format": "webp",
  "cfg_renorm_min": 1,
  "cfg_text_scale": 4,
  "output_quality": 90,
  "timestep_shift": 3,
  "cfg_renorm_type": "text_channel",
  "enable_thinking": false,
  "num_inference_steps": 50
}
```

### Example Output
The request returned an edited image and `text` set to `null`:

![BAGEL output: the same woman reading a folded newspaper on a modern London Underground train](https://blog.api.market/wp-content/uploads/2026/10/bagel-output-tv72rj0q-api-market.webp)

## Input Parameters
| Field | Type | Required | Default | Allowed values | Description |
| --- | --- | --- | --- | --- | --- |
| `prompt` | string | Yes |  |  | Text prompt for generation, editing, or understanding |
| `image` | string (URL) | No | none |  | Input image URL for the image-editing and image-understanding tasks |
| `task` | string | No | `text-to-image` | `text-to-image`, `image-editing`, `image-understanding` | Task to perform |
| `enable_thinking` | boolean | No | `false` |  | Enable chain-of-thought reasoning for better results |
| `cfg_text_scale` | number | No | `4` | 1 to 20 | Text guidance scale for how closely to follow the prompt |
| `cfg_img_scale` | number | No | `1.5` | 1 to 10 | Image guidance scale for preserving input image details |
| `num_inference_steps` | integer | No | `50` | 1 to 100 | Number of denoising steps |
| `timestep_shift` | number | No | `3` | 1 to 10 | How the denoising steps are split between composition and details |
| `cfg_renorm_type` | string | No | `global` | `global`, `local`, `text_channel` | CFG renormalization method |
| `cfg_renorm_min` | number | No | `1` | 0 to 1 | Minimum CFG renorm value |
| `seed` | integer | No | none |  | Random seed for reproducible results |
| `output_format` | string | No | `webp` | `webp`, `jpg`, `png` | Output image format |
| `output_quality` | integer | No | `90` | 1 to 100 | Image compression quality for lossy formats |

## Pricing
- **Pay As You Go (PAYG) pricing**: you only pay for what you use.
- **Price**: $0.055 per API unit.
- **1 API unit = 1 request with up to 50 seconds of processing time.** A request that finishes within 50 seconds costs one unit ($0.055).
- A request that runs longer bills one more unit for each started block of 50 seconds: more than 50 and up to 100 seconds is 2 units ($0.11), up to 150 seconds is 3 units ($0.165), and so on.
- **Example cost**: the example request took 70.7 seconds of processing, which is 2 units = **$0.11**.
- More `num_inference_steps` and `enable_thinking` take longer and can use more units.
- **Failed or cancelled predictions are not billed.**
- **GET requests to check prediction status** are free and don't count toward API unit usage.
- **No monthly subscription fee.**

## API Documentation

### Base URL
All API requests go to: `https://prod.api.market/api/v1/magicapi/unified-multimodal-image-ai-bytedance-bagel`

### Authentication
Send your api.market API key in the `x-api-market-key` header:
```
x-api-market-key: YOUR_API_MARKET_KEY
```

### Asynchronous Processing
**IMPORTANT**: This API is asynchronous. The flow is:
1. Send a POST request to `/predictions`. A successful request returns **201 Created** with a prediction `id` and a `status` of `starting` or `processing`.
2. Poll GET `/predictions/{id}` with that ID.
3. Keep polling while `status` is `starting` or `processing`.
4. Stop when `status` is `succeeded` (read `output`), `failed` or `canceled` (read `error`).

A polling interval of 2 seconds and an overall timeout of 600 seconds (10 minutes) work well.

### Endpoints

#### Create a Prediction
```bash
curl -X POST "https://prod.api.market/api/v1/magicapi/unified-multimodal-image-ai-bytedance-bagel/predictions" \
  -H "Content-Type: application/json" \
  -H "x-api-market-key: YOUR_API_MARKET_KEY" \
  -d '{
    "version": "7dd8def79e503990740db4704fa81af995d440fefe714958531d7044d2757c9c",
    "input": {
      "task": "image-editing",
      "image": "https://blog.api.market/wp-content/uploads/2026/10/bagel-input-api-market.png",
      "prompt": "She boards a modern London Tube, quietly reading a folded newspaper, wearing the same clothes",
      "cfg_img_scale": 2,
      "output_format": "webp",
      "cfg_renorm_min": 1,
      "cfg_text_scale": 4,
      "output_quality": 90,
      "timestep_shift": 3,
      "cfg_renorm_type": "text_channel",
      "enable_thinking": false,
      "num_inference_steps": 50
    }
  }'
```

Response (201 Created):
```json
{
  "id": "tv72rj0qexrmr0d1d0sstbks9r",
  "version": "7dd8def79e503990740db4704fa81af995d440fefe714958531d7044d2757c9c",
  "status": "starting",
  "created_at": "2026-10-23T09:59:01.435Z"
}
```

#### Check Prediction Status
```bash
curl -X GET "https://prod.api.market/api/v1/magicapi/unified-multimodal-image-ai-bytedance-bagel/predictions/{id}" \
  -H "x-api-market-key: YOUR_API_MARKET_KEY"
```

Response when the prediction has finished (200 OK):
```json
{
  "id": "tv72rj0qexrmr0d1d0sstbks9r",
  "status": "succeeded",
  "output": {
    "image": "https://blog.api.market/wp-content/uploads/2026/10/bagel-output-tv72rj0q-api-market.webp",
    "text": null
  },
  "error": null,
  "created_at": "2026-10-23T09:59:01.435Z",
  "started_at": "2026-10-23T10:00:27.712Z",
  "completed_at": "2026-10-23T10:01:38.427Z"
}
```

`output` is an object with two fields: `image`, the URL of the generated or edited image, and `text`, the text output (the answer for `image-understanding`; `null` in the example).

### Status Codes
- **201 Created**: prediction created (POST `/predictions`). This is the normal success response.
- **200 OK**: prediction details returned (GET `/predictions/{id}`).
- **400 Bad Request**: invalid input, for example a missing `prompt` or a `task` that is not in the list.
- **401 Unauthorized**: missing or invalid API key.
- **404 Not Found**: no prediction with that ID.
- **429 Too Many Requests**: rate limit exceeded.
- **500 Internal Server Error** or **503 Service Unavailable**: temporary server problem; retry later.

## Rate Limits
All plans allow 10 requests per second. Requests above that return `429 Too Many Requests`; back off before retrying and honour any `Retry-After` header. Polling a prediction with GET is free but still counts toward the per-second limit, so poll every 1 to 5 seconds rather than in a tight loop.

## Error Handling
- Check the HTTP status code of every request and read the response body for the error message.
- A prediction can also end with `status` set to `failed`; the `error` field then explains why. Failed predictions are not billed.
- Retry 429, 500 and 503 responses with exponential backoff. Fix the input before retrying a 400.

## Implementation Guide
### Example Code

#### JavaScript
```javascript
const apiKey = "YOUR_API_MARKET_KEY";
const baseUrl = "https://prod.api.market/api/v1/magicapi/unified-multimodal-image-ai-bytedance-bagel";

async function createAndWaitForPrediction(input) {
  const response = await fetch(`${baseUrl}/predictions`, {
    method: "POST",
    headers: {
      "Content-Type": "application/json",
      "x-api-market-key": apiKey
    },
    body: JSON.stringify({ version: "7dd8def79e503990740db4704fa81af995d440fefe714958531d7044d2757c9c", input })
  });

  if (response.status !== 201) {
    throw new Error(`API error: ${response.status} ${await response.text()}`);
  }

  const prediction = await response.json();
  return pollForResults(prediction.id);
}

const delay = (ms) => new Promise((resolve) => setTimeout(resolve, ms));

async function pollForResults(predictionId) {
  const maxTimeout = 600000;
  const pollInterval = 2000;
  const startTime = Date.now();

  while (Date.now() - startTime < maxTimeout) {
    const response = await fetch(`${baseUrl}/predictions/${predictionId}`, {
      headers: { "x-api-market-key": apiKey }
    });

    if (!response.ok) {
      throw new Error(`API error: ${response.status} ${await response.text()}`);
    }

    const prediction = await response.json();

    if (prediction.status === "succeeded") {
      return prediction.output;
    }

    if (prediction.status === "failed" || prediction.status === "canceled") {
      throw new Error(`Prediction ${prediction.status}: ${prediction.error || "Unknown error"}`);
    }

    await delay(pollInterval);
  }

  throw new Error("Prediction timed out after 10 minutes");
}

createAndWaitForPrediction({
  "task": "image-editing",
  "image": "https://blog.api.market/wp-content/uploads/2026/10/bagel-input-api-market.png",
  "prompt": "She boards a modern London Tube, quietly reading a folded newspaper, wearing the same clothes",
  "cfg_img_scale": 2,
  "output_format": "webp",
  "cfg_renorm_min": 1,
  "cfg_text_scale": 4,
  "output_quality": 90,
  "timestep_shift": 3,
  "cfg_renorm_type": "text_channel",
  "enable_thinking": false,
  "num_inference_steps": 50
}).then((output) => console.log(output));
```

#### Python
```python
import time
import requests

API_KEY = "YOUR_API_MARKET_KEY"
BASE_URL = "https://prod.api.market/api/v1/magicapi/unified-multimodal-image-ai-bytedance-bagel"


def create_and_wait_for_prediction(input_data):
    headers = {"Content-Type": "application/json", "x-api-market-key": API_KEY}
    response = requests.post(
        f"{BASE_URL}/predictions",
        headers=headers,
        json={"version": "7dd8def79e503990740db4704fa81af995d440fefe714958531d7044d2757c9c", "input": input_data},
    )

    if response.status_code != 201:
        raise Exception(f"API error: {response.status_code} {response.text}")

    return poll_for_results(response.json()["id"])


def poll_for_results(prediction_id):
    max_timeout = 600
    poll_interval = 2
    headers = {"x-api-market-key": API_KEY}
    start_time = time.time()

    while time.time() - start_time < max_timeout:
        response = requests.get(f"{BASE_URL}/predictions/{prediction_id}", headers=headers)

        if response.status_code != 200:
            raise Exception(f"API error: {response.status_code} {response.text}")

        prediction = response.json()

        if prediction["status"] == "succeeded":
            return prediction["output"]

        if prediction["status"] in ("failed", "canceled"):
            raise Exception(f"Prediction {prediction['status']}: {prediction.get('error') or 'Unknown error'}")

        time.sleep(poll_interval)

    raise Exception("Prediction timed out after 10 minutes")


output = create_and_wait_for_prediction({
    "task": "image-editing",
    "image": "https://blog.api.market/wp-content/uploads/2026/10/bagel-input-api-market.png",
    "prompt": "She boards a modern London Tube, quietly reading a folded newspaper, wearing the same clothes",
    "cfg_img_scale": 2,
    "output_format": "webp",
    "cfg_renorm_min": 1,
    "cfg_text_scale": 4,
    "output_quality": 90,
    "timestep_shift": 3,
    "cfg_renorm_type": "text_channel",
    "enable_thinking": False,
    "num_inference_steps": 50
})
print(output)
```

## FAQ
- **How do I edit an image?**
  - Set `task` to `image-editing`, pass the image as `image` and describe the change in `prompt`. Raise `cfg_img_scale` to keep more of the original.
- **How do I ask a question about an image?**
  - Set `task` to `image-understanding`, pass the `image` and put the question in `prompt`. Read the answer from `output.text`.
- **Do I need an image for text-to-image?**
  - No. `text-to-image` is the default task and only needs `prompt`.
- **What does `enable_thinking` do?**
  - It turns on chain-of-thought reasoning before the model produces its result. It is off by default.
- **What did the example request cost?**
  - 2 API units, $0.11, because it took 70.7 seconds of processing.
- **Is the 201 response from POST an error?**
  - No. 201 Created means the prediction was accepted. Use the returned `id` to poll GET `/predictions/{id}` until `status` is `succeeded`, `failed` or `canceled`.

## Contact Us
If you have any questions, reach out to us through the chat on the bottom right of the page. You can also email our support team at hello@magicapi.com.


Use ByteDance Seed's BAGEL to generate images from text, edit images with instructions and answer questions about images through one API.