Image model parameters
The parameters below are listed by exact model ID. Shared authentication and request examples are in the integration overview.
Reference images use the image field: file[] means multipart/form-data files, repeating the field for multiple images; a string type means JSON with image URLs or complete image Data URLs. Send recommended values explicitly. Shared examples are in the integration overview.
Checking available models…
GPT Image
Nano Banana
Grok Imagine
AWSD-5-pro-nsfw
| Field | Type | Operation | Required | Values and constraints | Recommended |
|---|---|---|---|---|---|
model | string | Generation / reference | Yes | AWSD-5-pro-nsfw; Exact public model ID. | Unspecified |
prompt | string | Generation / reference | Yes | Describe the image or requested edit. | Unspecified |
n | integer | Generation / reference | No | 1; Requested image count; additional images may increase the charge. | 1 |
size | string | Generation / reference | No | 1024x1024, 1152x864, 864x1152, 1424x800, 800x1424, 1248x832, 832x1248, 1568x672, 1536x1536, 1792x1344, 1344x1792, 2048x1152, 1152x2048, 1872x1248, 1248x1872, 2352x1008, 2048x2048, 2368x1776, 1776x2368, 2816x1584, 1584x2816, 2496x1664, 1664x2496, 3136x1344; Use a listed WIDTHxHEIGHT value. For automatic sizing, omit size; do not send the literal auto. | 2048x2048 |
output_format | string | Generation / reference | No | png, jpeg; Encoded image format; separate from the response container. | png |
response_format | string | Generation / reference | No | b64_json; Read every item in data[]. Decode b64_json or download url when present. | b64_json |
image | string | string[] | generation | Required for reference requests | max 10; Reference images as public URL or complete image Data URL strings. | Unspecified |
Sizes and aspect ratios
Choose size below. Resolution tiers and aspect ratios are size selectors, not separate request fields.
| Resolution tier | Aspect ratio | size |
|---|---|---|
1K | 1:1 | 1024x1024 |
1K | 4:3 | 1152x864 |
1K | 3:4 | 864x1152 |
1K | 16:9 | 1424x800 |
1K | 9:16 | 800x1424 |
1K | 3:2 | 1248x832 |
1K | 2:3 | 832x1248 |
1K | 21:9 | 1568x672 |
1.5K | 1:1 | 1536x1536 |
1.5K | 4:3 | 1792x1344 |
1.5K | 3:4 | 1344x1792 |
1.5K | 16:9 | 2048x1152 |
1.5K | 9:16 | 1152x2048 |
1.5K | 3:2 | 1872x1248 |
1.5K | 2:3 | 1248x1872 |
1.5K | 21:9 | 2352x1008 |
2K | 1:1 | 2048x2048 |
2K | 4:3 | 2368x1776 |
2K | 3:4 | 1776x2368 |
2K | 16:9 | 2816x1584 |
2K | 9:16 | 1584x2816 |
2K | 3:2 | 2496x1664 |
2K | 2:3 | 1664x2496 |
2K | 21:9 | 3136x1344 |
Results
- Read every item in
data[]. Decode b64_json or download url when present. - This guide documents only the fields and directives listed for each model; it does not define support for other fields such as mask, stream, or tools.
Midjourney v6
| Field | Type | Operation | Required | Values and constraints | Recommended |
|---|---|---|---|---|---|
model | string | Generation / reference | Yes | Midjourney v6; Exact public model ID. | Unspecified |
prompt | string | Generation / reference | Yes | Describe the image to generate. | Unspecified |
n | integer | Generation / reference | No | ≥ 1; ≤ 50; For one standard Midjourney generation, use n=1; the task returns four candidate images. The current published range is 1–50. | 1 |
mode | string | Generation / reference | No | text-to-image; Generation mode; use text-to-image. This is not a speed setting. | text-to-image |
size | string | Generation / reference | No | 1024x1024, 1536x864, 864x1536, 1152x864, 864x1152, 1248x832, 832x1248, 1024x1280, 1280x1024, 1344x576; Use a supported aspect-ratio string or listed WIDTHxHEIGHT value. Keep size, aspect_ratio, and resolution consistent; final pixel dimensions may differ. | 1024x1024 |
quality | string | Generation / reference | No | 0.25, 0.5, 1, 2; Quality values are 0.25, 0.5, 1, and 2; default 1. Quality is independent of resolution. | 1 |
resolution | string | Generation / reference | No | 1k; Use 1k for v6, v6.1, and v7. The 2k option is published for v8.1 and v8.2. | Unspecified |
aspect_ratio | string | Generation / reference | No | 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 4:5, 5:4, 21:9; Supported values: 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 4:5, 5:4, and 21:9. | Unspecified |
style | string | Generation / reference | No | ``, raw; MJ-specific control; its availability depends on the selected version. | Unspecified |
stylize | integer | Generation / reference | No | ≥ 0; ≤ 1000; MJ-specific control; its availability depends on the selected version. | 100 |
chaos | integer | Generation / reference | No | ≥ 0; ≤ 100; MJ-specific control; its availability depends on the selected version. | 0 |
seed | integer | Generation / reference | No | ≥ 0; ≤ 4294967295; MJ-specific control; its availability depends on the selected version. | Unspecified |
sref | string | Generation / reference | No | MJ-specific control; its availability depends on the selected version. | Unspecified |
sw | integer | Generation / reference | No | ≥ 0; ≤ 1000; MJ-specific control; its availability depends on the selected version. | 100 |
iw | number | Generation / reference | No | 0.5, 1, 2, 3; MJ-specific control; its availability depends on the selected version. | 2 |
tile | boolean | Generation / reference | No | MJ-specific control; its availability depends on the selected version. | false |
weird | integer | Generation / reference | No | ≥ 0; ≤ 3000; MJ-specific control; its availability depends on the selected version. | 0 |
negative_prompt | string | Generation / reference | No | MJ-specific control; its availability depends on the selected version. | Unspecified |
hd | boolean | Generation / reference | No | false, true; MJ-specific control; its availability depends on the selected version. | Unspecified |
draft | boolean | Generation / reference | No | false, true; MJ-specific control; its availability depends on the selected version. | Unspecified |
cref | string | Generation / reference | No | MJ-specific control; its availability depends on the selected version. | Unspecified |
cw | integer | Generation / reference | No | 0, 50, 100; ≥ 0; ≤ 100; MJ-specific control; its availability depends on the selected version. | Unspecified |
stop | integer | Generation / reference | No | 50, 80, 100; ≥ 10; ≤ 100; MJ-specific control; its availability depends on the selected version. | Unspecified |
bot_type | string | Generation / reference | No | MID_JOURNEY, NIJI_JOURNEY; MJ-specific control; its availability depends on the selected version. | MID_JOURNEY |
extra | string | Generation / reference | No | Additional MJ command-line options not represented by structured fields. Do not duplicate or conflict with structured parameters. | Unspecified |
count | integer | Generation / reference | No | ≥ 1; ≤ 50; count is an alias for n, with a range of 1–50. Use count=1 for one task, which returns four candidate images. | Unspecified |
output_format | string | Generation / reference | No | png, jpeg; Encoded image format; separate from the response container. | jpeg |
response_format | string | Generation / reference | No | url, b64_json; Read every item in data[]. Decode b64_json or download url when present. | Unspecified |
image | string | string[] | generation | No | max 1; For reference input, send either image or images, not both; the total limit is one image. | Unspecified |
images | ImageInputReference[] | generation | No | max 1; For reference input, send either image or images, not both; the total limit is one image. | Unspecified |
Sizes and aspect ratios
Choose size below. Resolution tiers and aspect ratios are size selectors, not separate request fields.
| Resolution tier | Aspect ratio | size |
|---|---|---|
1K | 1:1 | 1024x1024 |
1K | 16:9 | 1536x864 |
1K | 9:16 | 864x1536 |
1K | 4:3 | 1152x864 |
1K | 3:4 | 864x1152 |
1K | 3:2 | 1248x832 |
1K | 2:3 | 832x1248 |
1K | 4:5 | 1024x1280 |
1K | 5:4 | 1280x1024 |
1K | 21:9 | 1344x576 |
Request example
{
"model": "YOUR_MODEL",
"prompt": "A glass lighthouse above misty cliffs at dawn",
"n": 1,
"size": "16:9",
"aspect_ratio": "16:9",
"resolution": "1k",
"quality": "1",
"style": "raw",
"stylize": 100,
"chaos": 0,
"output_format": "jpeg",
"response_format": "url"
}
Results
- Send a reference as a public HTTPS URL or image Data URL in image or images. The current published model capability lists at most one reference image; a reference guides generation and does not guarantee an exact copy or edit.
- Use structured JSON fields such as size, aspect_ratio, resolution, and quality instead of putting --ar, --v, or --quality in prompt. The selected model determines the version. Use extra for additional MJ command-line options not represented by a structured field.
- With n=1, one Midjourney task returns four images. After completion, process every item in data[] and read url or b64_json according to response_format. No fixed result-URL expiry is stated here; save results you need to keep.
- Controls: style selects raw rendering; stylize controls stylization; chaos controls variation; seed fixes the random seed; sref and sw set a style reference and its weight; iw sets image-reference weight; tile enables seamless tiling; weird controls unusualness; negative_prompt excludes content. Use only the controls published for the selected version.
- Version-specific options: hd is for v8.1/v8.2; draft is for v7/v8.1; cref, cw, and stop are for v6/v6.1; NIJI_JOURNEY is available in bot_type for v6/v7. The version follows the selected model ID; omit version.
- Choose the exact Midjourney model ID exposed to your API key. The version is selected by model; do not use a prompt directive or version field to switch it.
- This guide documents only the fields and directives listed for each model; it does not define support for other fields such as mask, stream, or tools.
Midjourney v6.1
| Field | Type | Operation | Required | Values and constraints | Recommended |
|---|---|---|---|---|---|
model | string | Generation / reference | Yes | Midjourney v6.1; Exact public model ID. | Unspecified |
prompt | string | Generation / reference | Yes | Describe the image to generate. | Unspecified |
n | integer | Generation / reference | No | ≥ 1; ≤ 50; For one standard Midjourney generation, use n=1; the task returns four candidate images. The current published range is 1–50. | 1 |
mode | string | Generation / reference | No | text-to-image; Generation mode; use text-to-image. This is not a speed setting. | text-to-image |
size | string | Generation / reference | No | 1024x1024, 1536x864, 864x1536, 1152x864, 864x1152, 1248x832, 832x1248, 1024x1280, 1280x1024, 1344x576; Use a supported aspect-ratio string or listed WIDTHxHEIGHT value. Keep size, aspect_ratio, and resolution consistent; final pixel dimensions may differ. | 1024x1024 |
quality | string | Generation / reference | No | 0.25, 0.5, 1, 2; Quality values are 0.25, 0.5, 1, and 2; default 1. Quality is independent of resolution. | 1 |
resolution | string | Generation / reference | No | 1k; Use 1k for v6, v6.1, and v7. The 2k option is published for v8.1 and v8.2. | Unspecified |
aspect_ratio | string | Generation / reference | No | 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 4:5, 5:4, 21:9; Supported values: 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 4:5, 5:4, and 21:9. | Unspecified |
style | string | Generation / reference | No | ``, raw; MJ-specific control; its availability depends on the selected version. | Unspecified |
stylize | integer | Generation / reference | No | ≥ 0; ≤ 1000; MJ-specific control; its availability depends on the selected version. | 100 |
chaos | integer | Generation / reference | No | ≥ 0; ≤ 100; MJ-specific control; its availability depends on the selected version. | 0 |
seed | integer | Generation / reference | No | ≥ 0; ≤ 4294967295; MJ-specific control; its availability depends on the selected version. | Unspecified |
sref | string | Generation / reference | No | MJ-specific control; its availability depends on the selected version. | Unspecified |
sw | integer | Generation / reference | No | ≥ 0; ≤ 1000; MJ-specific control; its availability depends on the selected version. | 100 |
iw | number | Generation / reference | No | 0.5, 1, 2, 3; MJ-specific control; its availability depends on the selected version. | 2 |
tile | boolean | Generation / reference | No | MJ-specific control; its availability depends on the selected version. | false |
weird | integer | Generation / reference | No | ≥ 0; ≤ 3000; MJ-specific control; its availability depends on the selected version. | 0 |
negative_prompt | string | Generation / reference | No | MJ-specific control; its availability depends on the selected version. | Unspecified |
hd | boolean | Generation / reference | No | false, true; MJ-specific control; its availability depends on the selected version. | Unspecified |
draft | boolean | Generation / reference | No | false, true; MJ-specific control; its availability depends on the selected version. | Unspecified |
cref | string | Generation / reference | No | MJ-specific control; its availability depends on the selected version. | Unspecified |
cw | integer | Generation / reference | No | 0, 50, 100; ≥ 0; ≤ 100; MJ-specific control; its availability depends on the selected version. | Unspecified |
stop | integer | Generation / reference | No | 50, 80, 100; ≥ 10; ≤ 100; MJ-specific control; its availability depends on the selected version. | Unspecified |
bot_type | string | Generation / reference | No | MID_JOURNEY, NIJI_JOURNEY; MJ-specific control; its availability depends on the selected version. | MID_JOURNEY |
extra | string | Generation / reference | No | Additional MJ command-line options not represented by structured fields. Do not duplicate or conflict with structured parameters. | Unspecified |
count | integer | Generation / reference | No | ≥ 1; ≤ 50; count is an alias for n, with a range of 1–50. Use count=1 for one task, which returns four candidate images. | Unspecified |
output_format | string | Generation / reference | No | png, jpeg; Encoded image format; separate from the response container. | jpeg |
response_format | string | Generation / reference | No | url, b64_json; Read every item in data[]. Decode b64_json or download url when present. | Unspecified |
image | string | string[] | generation | No | max 1; For reference input, send either image or images, not both; the total limit is one image. | Unspecified |
images | ImageInputReference[] | generation | No | max 1; For reference input, send either image or images, not both; the total limit is one image. | Unspecified |
Sizes and aspect ratios
Choose size below. Resolution tiers and aspect ratios are size selectors, not separate request fields.
| Resolution tier | Aspect ratio | size |
|---|---|---|
1K | 1:1 | 1024x1024 |
1K | 16:9 | 1536x864 |
1K | 9:16 | 864x1536 |
1K | 4:3 | 1152x864 |
1K | 3:4 | 864x1152 |
1K | 3:2 | 1248x832 |
1K | 2:3 | 832x1248 |
1K | 4:5 | 1024x1280 |
1K | 5:4 | 1280x1024 |
1K | 21:9 | 1344x576 |
Request example
{
"model": "YOUR_MODEL",
"prompt": "A glass lighthouse above misty cliffs at dawn",
"n": 1,
"size": "16:9",
"aspect_ratio": "16:9",
"resolution": "1k",
"quality": "1",
"style": "raw",
"stylize": 100,
"chaos": 0,
"output_format": "jpeg",
"response_format": "url"
}
Results
- Send a reference as a public HTTPS URL or image Data URL in image or images. The current published model capability lists at most one reference image; a reference guides generation and does not guarantee an exact copy or edit.
- Use structured JSON fields such as size, aspect_ratio, resolution, and quality instead of putting --ar, --v, or --quality in prompt. The selected model determines the version. Use extra for additional MJ command-line options not represented by a structured field.
- With n=1, one Midjourney task returns four images. After completion, process every item in data[] and read url or b64_json according to response_format. No fixed result-URL expiry is stated here; save results you need to keep.
- Controls: style selects raw rendering; stylize controls stylization; chaos controls variation; seed fixes the random seed; sref and sw set a style reference and its weight; iw sets image-reference weight; tile enables seamless tiling; weird controls unusualness; negative_prompt excludes content. Use only the controls published for the selected version.
- Version-specific options: hd is for v8.1/v8.2; draft is for v7/v8.1; cref, cw, and stop are for v6/v6.1; NIJI_JOURNEY is available in bot_type for v6/v7. The version follows the selected model ID; omit version.
- Choose the exact Midjourney model ID exposed to your API key. The version is selected by model; do not use a prompt directive or version field to switch it.
- This guide documents only the fields and directives listed for each model; it does not define support for other fields such as mask, stream, or tools.
Midjourney v7
| Field | Type | Operation | Required | Values and constraints | Recommended |
|---|---|---|---|---|---|
model | string | Generation / reference | Yes | Midjourney v7; Exact public model ID. | Unspecified |
prompt | string | Generation / reference | Yes | Describe the image to generate. | Unspecified |
n | integer | Generation / reference | No | ≥ 1; ≤ 50; For one standard Midjourney generation, use n=1; the task returns four candidate images. The current published range is 1–50. | 1 |
mode | string | Generation / reference | No | text-to-image; Generation mode; use text-to-image. This is not a speed setting. | text-to-image |
size | string | Generation / reference | No | 1024x1024, 1536x864, 864x1536, 1152x864, 864x1152, 1248x832, 832x1248, 1024x1280, 1280x1024, 1344x576; Use a supported aspect-ratio string or listed WIDTHxHEIGHT value. Keep size, aspect_ratio, and resolution consistent; final pixel dimensions may differ. | 1024x1024 |
quality | string | Generation / reference | No | 0.25, 0.5, 1, 2; Quality values are 0.25, 0.5, 1, and 2; default 1. Quality is independent of resolution. | 1 |
resolution | string | Generation / reference | No | 1k; Use 1k for v6, v6.1, and v7. The 2k option is published for v8.1 and v8.2. | Unspecified |
aspect_ratio | string | Generation / reference | No | 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 4:5, 5:4, 21:9; Supported values: 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 4:5, 5:4, and 21:9. | Unspecified |
style | string | Generation / reference | No | ``, raw; MJ-specific control; its availability depends on the selected version. | Unspecified |
stylize | integer | Generation / reference | No | ≥ 0; ≤ 1000; MJ-specific control; its availability depends on the selected version. | 100 |
chaos | integer | Generation / reference | No | ≥ 0; ≤ 100; MJ-specific control; its availability depends on the selected version. | 0 |
seed | integer | Generation / reference | No | ≥ 0; ≤ 4294967295; MJ-specific control; its availability depends on the selected version. | Unspecified |
sref | string | Generation / reference | No | MJ-specific control; its availability depends on the selected version. | Unspecified |
sw | integer | Generation / reference | No | ≥ 0; ≤ 1000; MJ-specific control; its availability depends on the selected version. | 100 |
iw | number | Generation / reference | No | 0.5, 1, 2, 3; MJ-specific control; its availability depends on the selected version. | 2 |
tile | boolean | Generation / reference | No | MJ-specific control; its availability depends on the selected version. | false |
weird | integer | Generation / reference | No | ≥ 0; ≤ 3000; MJ-specific control; its availability depends on the selected version. | 0 |
negative_prompt | string | Generation / reference | No | MJ-specific control; its availability depends on the selected version. | Unspecified |
hd | boolean | Generation / reference | No | false, true; MJ-specific control; its availability depends on the selected version. | Unspecified |
draft | boolean | Generation / reference | No | false, true; MJ-specific control; its availability depends on the selected version. | Unspecified |
cref | string | Generation / reference | No | MJ-specific control; its availability depends on the selected version. | Unspecified |
cw | integer | Generation / reference | No | 0, 50, 100; ≥ 0; ≤ 100; MJ-specific control; its availability depends on the selected version. | Unspecified |
stop | integer | Generation / reference | No | 50, 80, 100; ≥ 10; ≤ 100; MJ-specific control; its availability depends on the selected version. | Unspecified |
bot_type | string | Generation / reference | No | MID_JOURNEY, NIJI_JOURNEY; MJ-specific control; its availability depends on the selected version. | MID_JOURNEY |
extra | string | Generation / reference | No | Additional MJ command-line options not represented by structured fields. Do not duplicate or conflict with structured parameters. | Unspecified |
count | integer | Generation / reference | No | ≥ 1; ≤ 50; count is an alias for n, with a range of 1–50. Use count=1 for one task, which returns four candidate images. | Unspecified |
output_format | string | Generation / reference | No | png, jpeg; Encoded image format; separate from the response container. | jpeg |
response_format | string | Generation / reference | No | url, b64_json; Read every item in data[]. Decode b64_json or download url when present. | Unspecified |
image | string | string[] | generation | No | max 1; For reference input, send either image or images, not both; the total limit is one image. | Unspecified |
images | ImageInputReference[] | generation | No | max 1; For reference input, send either image or images, not both; the total limit is one image. | Unspecified |
Sizes and aspect ratios
Choose size below. Resolution tiers and aspect ratios are size selectors, not separate request fields.
| Resolution tier | Aspect ratio | size |
|---|---|---|
1K | 1:1 | 1024x1024 |
1K | 16:9 | 1536x864 |
1K | 9:16 | 864x1536 |
1K | 4:3 | 1152x864 |
1K | 3:4 | 864x1152 |
1K | 3:2 | 1248x832 |
1K | 2:3 | 832x1248 |
1K | 4:5 | 1024x1280 |
1K | 5:4 | 1280x1024 |
1K | 21:9 | 1344x576 |
Request example
{
"model": "YOUR_MODEL",
"prompt": "A glass lighthouse above misty cliffs at dawn",
"n": 1,
"size": "16:9",
"aspect_ratio": "16:9",
"resolution": "1k",
"quality": "1",
"style": "raw",
"stylize": 100,
"chaos": 0,
"output_format": "jpeg",
"response_format": "url"
}
Results
- Send a reference as a public HTTPS URL or image Data URL in image or images. The current published model capability lists at most one reference image; a reference guides generation and does not guarantee an exact copy or edit.
- Use structured JSON fields such as size, aspect_ratio, resolution, and quality instead of putting --ar, --v, or --quality in prompt. The selected model determines the version. Use extra for additional MJ command-line options not represented by a structured field.
- With n=1, one Midjourney task returns four images. After completion, process every item in data[] and read url or b64_json according to response_format. No fixed result-URL expiry is stated here; save results you need to keep.
- Controls: style selects raw rendering; stylize controls stylization; chaos controls variation; seed fixes the random seed; sref and sw set a style reference and its weight; iw sets image-reference weight; tile enables seamless tiling; weird controls unusualness; negative_prompt excludes content. Use only the controls published for the selected version.
- Version-specific options: hd is for v8.1/v8.2; draft is for v7/v8.1; cref, cw, and stop are for v6/v6.1; NIJI_JOURNEY is available in bot_type for v6/v7. The version follows the selected model ID; omit version.
- Choose the exact Midjourney model ID exposed to your API key. The version is selected by model; do not use a prompt directive or version field to switch it.
- This guide documents only the fields and directives listed for each model; it does not define support for other fields such as mask, stream, or tools.
Midjourney v8.1
| Field | Type | Operation | Required | Values and constraints | Recommended |
|---|---|---|---|---|---|
model | string | Generation / reference | Yes | Midjourney v8.1; Exact public model ID. | Unspecified |
prompt | string | Generation / reference | Yes | Describe the image to generate. | Unspecified |
n | integer | Generation / reference | No | ≥ 1; ≤ 50; For one standard Midjourney generation, use n=1; the task returns four candidate images. The current published range is 1–50. | 1 |
mode | string | Generation / reference | No | text-to-image; Generation mode; use text-to-image. This is not a speed setting. | text-to-image |
size | string | Generation / reference | No | 1024x1024, 1536x864, 864x1536, 1152x864, 864x1152, 1248x832, 832x1248, 1024x1280, 1280x1024, 1344x576, 2048x2048, 2048x1152, 1152x2048, 2048x1536, 1536x2048, 2048x1360, 1360x2048, 1632x2040, 2040x1632, 2016x864; Use a supported aspect-ratio string or listed WIDTHxHEIGHT value. Keep size, aspect_ratio, and resolution consistent; final pixel dimensions may differ. | 1024x1024 |
quality | string | Generation / reference | No | 0.25, 0.5, 1, 2; Quality values are 0.25, 0.5, 1, and 2; default 1. Quality is independent of resolution. | 1 |
resolution | string | Generation / reference | No | 1k, 2k; Use 1k for v6, v6.1, and v7. The 2k option is published for v8.1 and v8.2. | Unspecified |
aspect_ratio | string | Generation / reference | No | 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 4:5, 5:4, 21:9; Supported values: 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 4:5, 5:4, and 21:9. | Unspecified |
style | string | Generation / reference | No | ``, raw; MJ-specific control; its availability depends on the selected version. | Unspecified |
stylize | integer | Generation / reference | No | ≥ 0; ≤ 1000; MJ-specific control; its availability depends on the selected version. | 100 |
chaos | integer | Generation / reference | No | ≥ 0; ≤ 100; MJ-specific control; its availability depends on the selected version. | 0 |
seed | integer | Generation / reference | No | ≥ 0; ≤ 4294967295; MJ-specific control; its availability depends on the selected version. | Unspecified |
sref | string | Generation / reference | No | MJ-specific control; its availability depends on the selected version. | Unspecified |
sw | integer | Generation / reference | No | ≥ 0; ≤ 1000; MJ-specific control; its availability depends on the selected version. | 100 |
iw | number | Generation / reference | No | 0.5, 1, 2, 3; MJ-specific control; its availability depends on the selected version. | 2 |
tile | boolean | Generation / reference | No | MJ-specific control; its availability depends on the selected version. | false |
weird | integer | Generation / reference | No | ≥ 0; ≤ 3000; MJ-specific control; its availability depends on the selected version. | 0 |
negative_prompt | string | Generation / reference | No | MJ-specific control; its availability depends on the selected version. | Unspecified |
hd | boolean | Generation / reference | No | false, true; MJ-specific control; its availability depends on the selected version. | Unspecified |
draft | boolean | Generation / reference | No | false, true; MJ-specific control; its availability depends on the selected version. | Unspecified |
cref | string | Generation / reference | No | MJ-specific control; its availability depends on the selected version. | Unspecified |
cw | integer | Generation / reference | No | 0, 50, 100; ≥ 0; ≤ 100; MJ-specific control; its availability depends on the selected version. | Unspecified |
stop | integer | Generation / reference | No | 50, 80, 100; ≥ 10; ≤ 100; MJ-specific control; its availability depends on the selected version. | Unspecified |
bot_type | string | Generation / reference | No | MID_JOURNEY, NIJI_JOURNEY; MJ-specific control; its availability depends on the selected version. | MID_JOURNEY |
extra | string | Generation / reference | No | Additional MJ command-line options not represented by structured fields. Do not duplicate or conflict with structured parameters. | Unspecified |
count | integer | Generation / reference | No | ≥ 1; ≤ 50; count is an alias for n, with a range of 1–50. Use count=1 for one task, which returns four candidate images. | Unspecified |
output_format | string | Generation / reference | No | png, jpeg; Encoded image format; separate from the response container. | jpeg |
response_format | string | Generation / reference | No | url, b64_json; Read every item in data[]. Decode b64_json or download url when present. | Unspecified |
image | string | string[] | generation | No | max 1; For reference input, send either image or images, not both; the total limit is one image. | Unspecified |
images | ImageInputReference[] | generation | No | max 1; For reference input, send either image or images, not both; the total limit is one image. | Unspecified |
Sizes and aspect ratios
Choose size below. Resolution tiers and aspect ratios are size selectors, not separate request fields.
| Resolution tier | Aspect ratio | size |
|---|---|---|
1K | 1:1 | 1024x1024 |
1K | 16:9 | 1536x864 |
1K | 9:16 | 864x1536 |
1K | 4:3 | 1152x864 |
1K | 3:4 | 864x1152 |
1K | 3:2 | 1248x832 |
1K | 2:3 | 832x1248 |
1K | 4:5 | 1024x1280 |
1K | 5:4 | 1280x1024 |
1K | 21:9 | 1344x576 |
2K | 1:1 | 2048x2048 |
2K | 16:9 | 2048x1152 |
2K | 9:16 | 1152x2048 |
2K | 4:3 | 2048x1536 |
2K | 3:4 | 1536x2048 |
2K | 3:2 | 2048x1360 |
2K | 2:3 | 1360x2048 |
2K | 4:5 | 1632x2040 |
2K | 5:4 | 2040x1632 |
2K | 21:9 | 2016x864 |
Request example
{
"model": "YOUR_MODEL",
"prompt": "A glass lighthouse above misty cliffs at dawn",
"n": 1,
"size": "16:9",
"aspect_ratio": "16:9",
"resolution": "1k",
"quality": "1",
"style": "raw",
"stylize": 100,
"chaos": 0,
"output_format": "jpeg",
"response_format": "url"
}
Results
- Send a reference as a public HTTPS URL or image Data URL in image or images. The current published model capability lists at most one reference image; a reference guides generation and does not guarantee an exact copy or edit.
- Use structured JSON fields such as size, aspect_ratio, resolution, and quality instead of putting --ar, --v, or --quality in prompt. The selected model determines the version. Use extra for additional MJ command-line options not represented by a structured field.
- With n=1, one Midjourney task returns four images. After completion, process every item in data[] and read url or b64_json according to response_format. No fixed result-URL expiry is stated here; save results you need to keep.
- Controls: style selects raw rendering; stylize controls stylization; chaos controls variation; seed fixes the random seed; sref and sw set a style reference and its weight; iw sets image-reference weight; tile enables seamless tiling; weird controls unusualness; negative_prompt excludes content. Use only the controls published for the selected version.
- Version-specific options: hd is for v8.1/v8.2; draft is for v7/v8.1; cref, cw, and stop are for v6/v6.1; NIJI_JOURNEY is available in bot_type for v6/v7. The version follows the selected model ID; omit version.
- Choose the exact Midjourney model ID exposed to your API key. The version is selected by model; do not use a prompt directive or version field to switch it.
- This guide documents only the fields and directives listed for each model; it does not define support for other fields such as mask, stream, or tools.
Midjourney v8.2
| Field | Type | Operation | Required | Values and constraints | Recommended |
|---|---|---|---|---|---|
model | string | Generation / reference | Yes | Midjourney v8.2; Exact public model ID. | Unspecified |
prompt | string | Generation / reference | Yes | Describe the image to generate. | Unspecified |
n | integer | Generation / reference | No | ≥ 1; ≤ 50; For one standard Midjourney generation, use n=1; the task returns four candidate images. The current published range is 1–50. | 1 |
mode | string | Generation / reference | No | text-to-image; Generation mode; use text-to-image. This is not a speed setting. | text-to-image |
size | string | Generation / reference | No | 1024x1024, 1536x864, 864x1536, 1152x864, 864x1152, 1248x832, 832x1248, 1024x1280, 1280x1024, 1344x576, 2048x2048, 2048x1152, 1152x2048, 2048x1536, 1536x2048, 2048x1360, 1360x2048, 1632x2040, 2040x1632, 2016x864; Use a supported aspect-ratio string or listed WIDTHxHEIGHT value. Keep size, aspect_ratio, and resolution consistent; final pixel dimensions may differ. | 1024x1024 |
quality | string | Generation / reference | No | 0.25, 0.5, 1, 2; Quality values are 0.25, 0.5, 1, and 2; default 1. Quality is independent of resolution. | 1 |
resolution | string | Generation / reference | No | 1k, 2k; Use 1k for v6, v6.1, and v7. The 2k option is published for v8.1 and v8.2. | Unspecified |
aspect_ratio | string | Generation / reference | No | 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 4:5, 5:4, 21:9; Supported values: 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 4:5, 5:4, and 21:9. | Unspecified |
style | string | Generation / reference | No | ``, raw; MJ-specific control; its availability depends on the selected version. | Unspecified |
stylize | integer | Generation / reference | No | ≥ 0; ≤ 1000; MJ-specific control; its availability depends on the selected version. | 100 |
chaos | integer | Generation / reference | No | ≥ 0; ≤ 100; MJ-specific control; its availability depends on the selected version. | 0 |
seed | integer | Generation / reference | No | ≥ 0; ≤ 4294967295; MJ-specific control; its availability depends on the selected version. | Unspecified |
sref | string | Generation / reference | No | MJ-specific control; its availability depends on the selected version. | Unspecified |
sw | integer | Generation / reference | No | ≥ 0; ≤ 1000; MJ-specific control; its availability depends on the selected version. | 100 |
iw | number | Generation / reference | No | 0.5, 1, 2, 3; MJ-specific control; its availability depends on the selected version. | 2 |
tile | boolean | Generation / reference | No | MJ-specific control; its availability depends on the selected version. | false |
weird | integer | Generation / reference | No | ≥ 0; ≤ 3000; MJ-specific control; its availability depends on the selected version. | 0 |
negative_prompt | string | Generation / reference | No | MJ-specific control; its availability depends on the selected version. | Unspecified |
hd | boolean | Generation / reference | No | false, true; MJ-specific control; its availability depends on the selected version. | Unspecified |
draft | boolean | Generation / reference | No | false, true; MJ-specific control; its availability depends on the selected version. | Unspecified |
cref | string | Generation / reference | No | MJ-specific control; its availability depends on the selected version. | Unspecified |
cw | integer | Generation / reference | No | 0, 50, 100; ≥ 0; ≤ 100; MJ-specific control; its availability depends on the selected version. | Unspecified |
stop | integer | Generation / reference | No | 50, 80, 100; ≥ 10; ≤ 100; MJ-specific control; its availability depends on the selected version. | Unspecified |
bot_type | string | Generation / reference | No | MID_JOURNEY, NIJI_JOURNEY; MJ-specific control; its availability depends on the selected version. | MID_JOURNEY |
extra | string | Generation / reference | No | Additional MJ command-line options not represented by structured fields. Do not duplicate or conflict with structured parameters. | Unspecified |
count | integer | Generation / reference | No | ≥ 1; ≤ 50; count is an alias for n, with a range of 1–50. Use count=1 for one task, which returns four candidate images. | Unspecified |
output_format | string | Generation / reference | No | png, jpeg; Encoded image format; separate from the response container. | jpeg |
response_format | string | Generation / reference | No | url, b64_json; Read every item in data[]. Decode b64_json or download url when present. | Unspecified |
image | string | string[] | generation | No | max 1; For reference input, send either image or images, not both; the total limit is one image. | Unspecified |
images | ImageInputReference[] | generation | No | max 1; For reference input, send either image or images, not both; the total limit is one image. | Unspecified |
Sizes and aspect ratios
Choose size below. Resolution tiers and aspect ratios are size selectors, not separate request fields.
| Resolution tier | Aspect ratio | size |
|---|---|---|
1K | 1:1 | 1024x1024 |
1K | 16:9 | 1536x864 |
1K | 9:16 | 864x1536 |
1K | 4:3 | 1152x864 |
1K | 3:4 | 864x1152 |
1K | 3:2 | 1248x832 |
1K | 2:3 | 832x1248 |
1K | 4:5 | 1024x1280 |
1K | 5:4 | 1280x1024 |
1K | 21:9 | 1344x576 |
2K | 1:1 | 2048x2048 |
2K | 16:9 | 2048x1152 |
2K | 9:16 | 1152x2048 |
2K | 4:3 | 2048x1536 |
2K | 3:4 | 1536x2048 |
2K | 3:2 | 2048x1360 |
2K | 2:3 | 1360x2048 |
2K | 4:5 | 1632x2040 |
2K | 5:4 | 2040x1632 |
2K | 21:9 | 2016x864 |
Request example
{
"model": "YOUR_MODEL",
"prompt": "A glass lighthouse above misty cliffs at dawn",
"n": 1,
"size": "16:9",
"aspect_ratio": "16:9",
"resolution": "1k",
"quality": "1",
"style": "raw",
"stylize": 100,
"chaos": 0,
"output_format": "jpeg",
"response_format": "url"
}
Results
- Send a reference as a public HTTPS URL or image Data URL in image or images. The current published model capability lists at most one reference image; a reference guides generation and does not guarantee an exact copy or edit.
- Use structured JSON fields such as size, aspect_ratio, resolution, and quality instead of putting --ar, --v, or --quality in prompt. The selected model determines the version. Use extra for additional MJ command-line options not represented by a structured field.
- With n=1, one Midjourney task returns four images. After completion, process every item in data[] and read url or b64_json according to response_format. No fixed result-URL expiry is stated here; save results you need to keep.
- Controls: style selects raw rendering; stylize controls stylization; chaos controls variation; seed fixes the random seed; sref and sw set a style reference and its weight; iw sets image-reference weight; tile enables seamless tiling; weird controls unusualness; negative_prompt excludes content. Use only the controls published for the selected version.
- Version-specific options: hd is for v8.1/v8.2; draft is for v7/v8.1; cref, cw, and stop are for v6/v6.1; NIJI_JOURNEY is available in bot_type for v6/v7. The version follows the selected model ID; omit version.
- Choose the exact Midjourney model ID exposed to your API key. The version is selected by model; do not use a prompt directive or version field to switch it.
- This guide documents only the fields and directives listed for each model; it does not define support for other fields such as mask, stream, or tools.
gpt-image-2
| Field | Type | Operation | Required | Values and constraints | Recommended |
|---|---|---|---|---|---|
model | string | Generation / reference | Yes | gpt-image-2; Exact public model ID. | Unspecified |
prompt | string | Generation / reference | Yes | Describe the image or requested edit. | Unspecified |
n | integer | Generation / reference | No | 1, 2, 3, 4, 5, 6, 7, 8, 9, 10; Requested image count; additional images may increase the charge. | 1 |
size | string | Generation / reference | No | 1024x1024, 1536x1024, 1024x1536, 1152x864, 864x1152, 1536x864, 864x1536, 2048x2048, 2496x1664, 1664x2496, 2304x1728, 1728x2304, 2048x1152, 1152x2048, 4096x4096, 4096x2736, 2736x4096, 4096x3072, 3072x4096, 4096x2304, 2304x4096; Use a listed WIDTHxHEIGHT value. For automatic sizing, omit size; do not send the literal auto. | 1024x1024 |
quality | string | Generation / reference | No | auto, low, medium, high; Use only the values listed for this exact model. | medium |
output_format | string | Generation / reference | No | png, jpeg, webp; Encoded image format; separate from the response container. | png |
response_format | string | Generation / reference | No | b64_json; Read every item in data[]. Decode b64_json or download url when present. | b64_json |
image | file[] | edit | Required for reference requests | max 16; Reference images as multipart files. Repeat image for multiple files. | Unspecified |
Sizes and aspect ratios
Choose size below. Resolution tiers and aspect ratios are size selectors, not separate request fields.
| Resolution tier | Aspect ratio | size |
|---|---|---|
1K | 1:1 | 1024x1024 |
1K | 3:2 | 1536x1024 |
1K | 2:3 | 1024x1536 |
1K | 4:3 | 1152x864 |
1K | 3:4 | 864x1152 |
1K | 16:9 | 1536x864 |
1K | 9:16 | 864x1536 |
2K | 1:1 | 2048x2048 |
2K | 3:2 | 2496x1664 |
2K | 2:3 | 1664x2496 |
2K | 4:3 | 2304x1728 |
2K | 3:4 | 1728x2304 |
2K | 16:9 | 2048x1152 |
2K | 9:16 | 1152x2048 |
4K | 1:1 | 4096x4096 |
4K | 3:2 | 4096x2736 |
4K | 2:3 | 2736x4096 |
4K | 4:3 | 4096x3072 |
4K | 3:4 | 3072x4096 |
4K | 16:9 | 4096x2304 |
4K | 9:16 | 2304x4096 |
Results
- Read every item in
data[]. Decode b64_json or download url when present. - This guide documents only the fields and directives listed for each model; it does not define support for other fields such as mask, stream, or tools.
gpt-image-2 4K
| Field | Type | Operation | Required | Values and constraints | Recommended |
|---|---|---|---|---|---|
model | string | Generation / reference | Yes | gpt-image-2 4K; Exact public model ID. | Unspecified |
prompt | string | Generation / reference | Yes | Describe the image or requested edit. | Unspecified |
n | integer | Generation / reference | No | 1, 2, 3, 4, 5, 6, 7, 8, 9, 10; Requested image count; additional images may increase the charge. | 1 |
size | string | Generation / reference | No | 4096x4096, 4096x2736, 2736x4096, 4096x3072, 3072x4096, 4096x2304, 2304x4096; Use a listed WIDTHxHEIGHT value. For automatic sizing, omit size; do not send the literal auto. | 4096x2304 |
quality | string | Generation / reference | No | low, medium, high, xhigh, max; Use only the values listed for this exact model. | medium |
output_format | string | Generation / reference | No | png, jpeg, webp; Encoded image format; separate from the response container. | png |
response_format | string | Generation / reference | No | b64_json; Read every item in data[]. Decode b64_json or download url when present. | b64_json |
image | file[] | edit | Required for reference requests | max 16; Reference images as multipart files. Repeat image for multiple files. | Unspecified |
Sizes and aspect ratios
Choose size below. Resolution tiers and aspect ratios are size selectors, not separate request fields.
| Resolution tier | Aspect ratio | size |
|---|---|---|
4K | 1:1 | 4096x4096 |
4K | 3:2 | 4096x2736 |
4K | 2:3 | 2736x4096 |
4K | 4:3 | 4096x3072 |
4K | 3:4 | 3072x4096 |
4K | 16:9 | 4096x2304 |
4K | 9:16 | 2304x4096 |
Results
- Read every item in
data[]. Decode b64_json or download url when present. - This guide documents only the fields and directives listed for each model; it does not define support for other fields such as mask, stream, or tools.
gpt-image-2 Adobe
| Field | Type | Operation | Required | Values and constraints | Recommended |
|---|---|---|---|---|---|
model | string | Generation / reference | Yes | gpt-image-2 Adobe; Exact public model ID. | Unspecified |
prompt | string | Generation / reference | Yes | Describe the image or requested edit. | Unspecified |
n | integer | Generation / reference | No | 1, 2, 3, 4, 5, 6, 7, 8, 9, 10; Requested image count; additional images may increase the charge. | 1 |
size | string | Generation / reference | No | 1024x1024, 1536x1024, 1024x1536, 1152x864, 864x1152, 1536x864, 864x1536, 2048x2048, 2496x1664, 1664x2496, 2304x1728, 1728x2304, 2048x1152, 1152x2048, 4096x4096, 4096x2736, 2736x4096, 4096x3072, 3072x4096, 4096x2304, 2304x4096; Use a listed WIDTHxHEIGHT value. For automatic sizing, omit size; do not send the literal auto. | 1536x1024 |
quality | string | Generation / reference | No | low, medium, high; Use only the values listed for this exact model. | high |
output_format | string | Generation / reference | No | png, jpeg, webp; Encoded image format; separate from the response container. | png |
response_format | string | Generation / reference | No | b64_json; Read every item in data[]. Decode b64_json or download url when present. | b64_json |
image | file[] | edit | Required for reference requests | max 16; Reference images as multipart files. Repeat image for multiple files. | Unspecified |
Results
- Read every item in
data[]. Decode b64_json or download url when present. - This guide documents only the fields and directives listed for each model; it does not define support for other fields such as mask, stream, or tools.
gpt-image-2.5
| Field | Type | Operation | Required | Values and constraints | Recommended |
|---|---|---|---|---|---|
model | string | Generation / reference | Yes | gpt-image-2.5; Exact public model ID. | Unspecified |
prompt | string | Generation / reference | Yes | Describe the image or requested edit. | Unspecified |
n | integer | Generation / reference | No | 1, 2, 3, 4, 5, 6, 7, 8, 9, 10; Requested image count; additional images may increase the charge. | 1 |
size | string | Generation / reference | No | 1024x1024, 1536x1024, 1024x1536, 1152x864, 864x1152, 1536x864, 864x1536, 2048x2048, 2496x1664, 1664x2496, 2304x1728, 1728x2304, 2048x1152, 1152x2048, 4096x4096, 4096x2736, 2736x4096, 4096x3072, 3072x4096, 4096x2304, 2304x4096; Use a listed WIDTHxHEIGHT value. For automatic sizing, omit size; do not send the literal auto. | 1024x1024 |
quality | string | Generation / reference | No | auto, low, medium, high; Use only the values listed for this exact model. | medium |
output_format | string | Generation / reference | No | png, jpeg, webp; Encoded image format; separate from the response container. | png |
response_format | string | Generation / reference | No | b64_json; Read every item in data[]. Decode b64_json or download url when present. | b64_json |
image | file[] | edit | Required for reference requests | max 16; Reference images as multipart files. Repeat image for multiple files. | Unspecified |
Sizes and aspect ratios
Choose size below. Resolution tiers and aspect ratios are size selectors, not separate request fields.
| Resolution tier | Aspect ratio | size |
|---|---|---|
1K | 1:1 | 1024x1024 |
1K | 3:2 | 1536x1024 |
1K | 2:3 | 1024x1536 |
1K | 4:3 | 1152x864 |
1K | 3:4 | 864x1152 |
1K | 16:9 | 1536x864 |
1K | 9:16 | 864x1536 |
2K | 1:1 | 2048x2048 |
2K | 3:2 | 2496x1664 |
2K | 2:3 | 1664x2496 |
2K | 4:3 | 2304x1728 |
2K | 3:4 | 1728x2304 |
2K | 16:9 | 2048x1152 |
2K | 9:16 | 1152x2048 |
4K | 1:1 | 4096x4096 |
4K | 3:2 | 4096x2736 |
4K | 2:3 | 2736x4096 |
4K | 4:3 | 4096x3072 |
4K | 3:4 | 3072x4096 |
4K | 16:9 | 4096x2304 |
4K | 9:16 | 2304x4096 |
Results
- Read every item in
data[]. Decode b64_json or download url when present. - This guide documents only the fields and directives listed for each model; it does not define support for other fields such as mask, stream, or tools.
gpt-image-2.5 Adobe
| Field | Type | Operation | Required | Values and constraints | Recommended |
|---|---|---|---|---|---|
model | string | Generation / reference | Yes | gpt-image-2.5 Adobe; Exact public model ID. | Unspecified |
prompt | string | Generation / reference | Yes | Describe the image or requested edit. | Unspecified |
n | integer | Generation / reference | No | 1, 2, 3, 4, 5, 6, 7, 8, 9, 10; Requested image count; additional images may increase the charge. | 1 |
size | string | Generation / reference | No | 1024x1024, 1536x1024, 1024x1536, 1152x864, 864x1152, 1536x864, 864x1536, 2048x2048, 2496x1664, 1664x2496, 2304x1728, 1728x2304, 2048x1152, 1152x2048, 4096x4096, 4096x2736, 2736x4096, 4096x3072, 3072x4096, 4096x2304, 2304x4096; Use a listed WIDTHxHEIGHT value. For automatic sizing, omit size; do not send the literal auto. | 1536x1024 |
quality | string | Generation / reference | No | low, medium, high; Use only the values listed for this exact model. | high |
output_format | string | Generation / reference | No | png, jpeg, webp; Encoded image format; separate from the response container. | png |
response_format | string | Generation / reference | No | b64_json; Read every item in data[]. Decode b64_json or download url when present. | b64_json |
image | file[] | edit | Required for reference requests | max 16; Reference images as multipart files. Repeat image for multiple files. | Unspecified |
Results
- Read every item in
data[]. Decode b64_json or download url when present. - This guide documents only the fields and directives listed for each model; it does not define support for other fields such as mask, stream, or tools.
gpt-image-2.5-flare
| Field | Type | Operation | Required | Values and constraints | Recommended |
|---|---|---|---|---|---|
model | string | Generation / reference | Yes | gpt-image-2.5-flare; Exact public model ID. | Unspecified |
prompt | string | Generation / reference | Yes | Describe the image or requested edit. | Unspecified |
n | integer | Generation / reference | No | 1, 2, 3, 4, 5, 6, 7, 8, 9, 10; Requested image count; additional images may increase the charge. | 1 |
size | string | Generation / reference | No | 1024x1024, 1536x1024, 1024x1536, 1152x864, 864x1152, 1536x864, 864x1536, 2048x2048, 2496x1664, 1664x2496, 2304x1728, 1728x2304, 2048x1152, 1152x2048, 4096x4096, 4096x2736, 2736x4096, 4096x3072, 3072x4096, 4096x2304, 2304x4096; Use a listed WIDTHxHEIGHT value. For automatic sizing, omit size; do not send the literal auto. | 1024x1024 |
quality | string | Generation / reference | No | auto, low, medium, high; Use only the values listed for this exact model. | medium |
output_format | string | Generation / reference | No | png, jpeg, webp; Encoded image format; separate from the response container. | png |
response_format | string | Generation / reference | No | b64_json; Read every item in data[]. Decode b64_json or download url when present. | b64_json |
image | file[] | edit | Required for reference requests | max 16; Reference images as multipart files. Repeat image for multiple files. | Unspecified |
Sizes and aspect ratios
Choose size below. Resolution tiers and aspect ratios are size selectors, not separate request fields.
| Resolution tier | Aspect ratio | size |
|---|---|---|
1K | 1:1 | 1024x1024 |
1K | 3:2 | 1536x1024 |
1K | 2:3 | 1024x1536 |
1K | 4:3 | 1152x864 |
1K | 3:4 | 864x1152 |
1K | 16:9 | 1536x864 |
1K | 9:16 | 864x1536 |
2K | 1:1 | 2048x2048 |
2K | 3:2 | 2496x1664 |
2K | 2:3 | 1664x2496 |
2K | 4:3 | 2304x1728 |
2K | 3:4 | 1728x2304 |
2K | 16:9 | 2048x1152 |
2K | 9:16 | 1152x2048 |
4K | 1:1 | 4096x4096 |
4K | 3:2 | 4096x2736 |
4K | 2:3 | 2736x4096 |
4K | 4:3 | 4096x3072 |
4K | 3:4 | 3072x4096 |
4K | 16:9 | 4096x2304 |
4K | 9:16 | 2304x4096 |
Results
- Read every item in
data[]. Decode b64_json or download url when present. - This guide documents only the fields and directives listed for each model; it does not define support for other fields such as mask, stream, or tools.
gpt-image-2.5-flare 4K
| Field | Type | Operation | Required | Values and constraints | Recommended |
|---|---|---|---|---|---|
model | string | Generation / reference | Yes | gpt-image-2.5-flare 4K; Exact public model ID. | Unspecified |
prompt | string | Generation / reference | Yes | Describe the image or requested edit. | Unspecified |
n | integer | Generation / reference | No | 1, 2, 3, 4, 5, 6, 7, 8, 9, 10; Requested image count; additional images may increase the charge. | 1 |
size | string | Generation / reference | No | 4096x4096, 4096x2736, 2736x4096, 4096x3072, 3072x4096, 4096x2304, 2304x4096; Use a listed WIDTHxHEIGHT value. For automatic sizing, omit size; do not send the literal auto. | 4096x2304 |
quality | string | Generation / reference | No | low, medium, high, xhigh, max; Use only the values listed for this exact model. | medium |
output_format | string | Generation / reference | No | png, jpeg, webp; Encoded image format; separate from the response container. | png |
response_format | string | Generation / reference | No | b64_json; Read every item in data[]. Decode b64_json or download url when present. | b64_json |
image | file[] | edit | Required for reference requests | max 16; Reference images as multipart files. Repeat image for multiple files. | Unspecified |
Sizes and aspect ratios
Choose size below. Resolution tiers and aspect ratios are size selectors, not separate request fields.
| Resolution tier | Aspect ratio | size |
|---|---|---|
4K | 1:1 | 4096x4096 |
4K | 3:2 | 4096x2736 |
4K | 2:3 | 2736x4096 |
4K | 4:3 | 4096x3072 |
4K | 3:4 | 3072x4096 |
4K | 16:9 | 4096x2304 |
4K | 9:16 | 2304x4096 |
Results
- Read every item in
data[]. Decode b64_json or download url when present. - This guide documents only the fields and directives listed for each model; it does not define support for other fields such as mask, stream, or tools.
gpt-image-2.5-flare Adobe
| Field | Type | Operation | Required | Values and constraints | Recommended |
|---|---|---|---|---|---|
model | string | Generation / reference | Yes | gpt-image-2.5-flare Adobe; Exact public model ID. | Unspecified |
prompt | string | Generation / reference | Yes | Describe the image or requested edit. | Unspecified |
n | integer | Generation / reference | No | 1, 2, 3, 4, 5, 6, 7, 8, 9, 10; Requested image count; additional images may increase the charge. | 1 |
size | string | Generation / reference | No | 1024x1024, 1536x1024, 1024x1536, 1152x864, 864x1152, 1536x864, 864x1536, 2048x2048, 2496x1664, 1664x2496, 2304x1728, 1728x2304, 2048x1152, 1152x2048, 4096x4096, 4096x2736, 2736x4096, 4096x3072, 3072x4096, 4096x2304, 2304x4096; Use a listed WIDTHxHEIGHT value. For automatic sizing, omit size; do not send the literal auto. | 1536x1024 |
quality | string | Generation / reference | No | low, medium, high; Use only the values listed for this exact model. | high |
output_format | string | Generation / reference | No | png, jpeg, webp; Encoded image format; separate from the response container. | png |
response_format | string | Generation / reference | No | b64_json; Read every item in data[]. Decode b64_json or download url when present. | b64_json |
image | file[] | edit | Required for reference requests | max 16; Reference images as multipart files. Repeat image for multiple files. | Unspecified |
Results
- Read every item in
data[]. Decode b64_json or download url when present. - This guide documents only the fields and directives listed for each model; it does not define support for other fields such as mask, stream, or tools.
gpt-image-2.5-flare Max
| Field | Type | Operation | Required | Values and constraints | Recommended |
|---|---|---|---|---|---|
model | string | Generation / reference | Yes | gpt-image-2.5-flare Max; Exact public model ID. | Unspecified |
prompt | string | Generation / reference | Yes | Describe the image or requested edit. | Unspecified |
n | integer | Generation / reference | No | 1, 2, 3, 4, 5, 6, 7, 8, 9, 10; Requested image count; additional images may increase the charge. | 1 |
size | string | Generation / reference | No | 1024x1024, 1536x1024, 1024x1536, 1152x864, 864x1152, 1536x864, 864x1536, 2048x2048, 2496x1664, 1664x2496, 2304x1728, 1728x2304, 2048x1152, 1152x2048, 4096x4096, 4096x2736, 2736x4096, 4096x3072, 3072x4096, 4096x2304, 2304x4096; Use a listed WIDTHxHEIGHT value. For automatic sizing, omit size; do not send the literal auto. | 4096x2736 |
quality | string | Generation / reference | No | auto, low, medium, high, xhigh, max; Use only the values listed for this exact model. | high |
output_format | string | Generation / reference | No | png, jpeg, webp; Encoded image format; separate from the response container. | png |
response_format | string | Generation / reference | No | b64_json; Read every item in data[]. Decode b64_json or download url when present. | b64_json |
image | file[] | edit | Required for reference requests | max 16; Reference images as multipart files. Repeat image for multiple files. | Unspecified |
Results
- Read every item in
data[]. Decode b64_json or download url when present. - This guide documents only the fields and directives listed for each model; it does not define support for other fields such as mask, stream, or tools.
gpt-image-2.5-sunburst
| Field | Type | Operation | Required | Values and constraints | Recommended |
|---|---|---|---|---|---|
model | string | Generation / reference | Yes | gpt-image-2.5-sunburst; Exact public model ID. | Unspecified |
prompt | string | Generation / reference | Yes | Describe the image or requested edit. | Unspecified |
n | integer | Generation / reference | No | 1, 2, 3, 4, 5, 6, 7, 8, 9, 10; Requested image count; additional images may increase the charge. | 1 |
size | string | Generation / reference | No | 1024x1024, 1536x1024, 1024x1536, 1152x864, 864x1152, 1536x864, 864x1536, 2048x2048, 2496x1664, 1664x2496, 2304x1728, 1728x2304, 2048x1152, 1152x2048, 4096x4096, 4096x2736, 2736x4096, 4096x3072, 3072x4096, 4096x2304, 2304x4096; Use a listed WIDTHxHEIGHT value. For automatic sizing, omit size; do not send the literal auto. | 1024x1024 |
quality | string | Generation / reference | No | auto, low, medium, high; Use only the values listed for this exact model. | medium |
output_format | string | Generation / reference | No | png, jpeg, webp; Encoded image format; separate from the response container. | png |
response_format | string | Generation / reference | No | b64_json; Read every item in data[]. Decode b64_json or download url when present. | b64_json |
image | file[] | edit | Required for reference requests | max 16; Reference images as multipart files. Repeat image for multiple files. | Unspecified |
Sizes and aspect ratios
Choose size below. Resolution tiers and aspect ratios are size selectors, not separate request fields.
| Resolution tier | Aspect ratio | size |
|---|---|---|
1K | 1:1 | 1024x1024 |
1K | 3:2 | 1536x1024 |
1K | 2:3 | 1024x1536 |
1K | 4:3 | 1152x864 |
1K | 3:4 | 864x1152 |
1K | 16:9 | 1536x864 |
1K | 9:16 | 864x1536 |
2K | 1:1 | 2048x2048 |
2K | 3:2 | 2496x1664 |
2K | 2:3 | 1664x2496 |
2K | 4:3 | 2304x1728 |
2K | 3:4 | 1728x2304 |
2K | 16:9 | 2048x1152 |
2K | 9:16 | 1152x2048 |
4K | 1:1 | 4096x4096 |
4K | 3:2 | 4096x2736 |
4K | 2:3 | 2736x4096 |
4K | 4:3 | 4096x3072 |
4K | 3:4 | 3072x4096 |
4K | 16:9 | 4096x2304 |
4K | 9:16 | 2304x4096 |
Results
- Read every item in
data[]. Decode b64_json or download url when present. - This guide documents only the fields and directives listed for each model; it does not define support for other fields such as mask, stream, or tools.
gpt-image-2.5-sunburst 4K
| Field | Type | Operation | Required | Values and constraints | Recommended |
|---|---|---|---|---|---|
model | string | Generation / reference | Yes | gpt-image-2.5-sunburst 4K; Exact public model ID. | Unspecified |
prompt | string | Generation / reference | Yes | Describe the image or requested edit. | Unspecified |
n | integer | Generation / reference | No | 1, 2, 3, 4, 5, 6, 7, 8, 9, 10; Requested image count; additional images may increase the charge. | 1 |
size | string | Generation / reference | No | 4096x4096, 4096x2736, 2736x4096, 4096x3072, 3072x4096, 4096x2304, 2304x4096; Use a listed WIDTHxHEIGHT value. For automatic sizing, omit size; do not send the literal auto. | 4096x2304 |
quality | string | Generation / reference | No | low, medium, high, xhigh, max; Use only the values listed for this exact model. | medium |
output_format | string | Generation / reference | No | png, jpeg, webp; Encoded image format; separate from the response container. | png |
response_format | string | Generation / reference | No | b64_json; Read every item in data[]. Decode b64_json or download url when present. | b64_json |
image | file[] | edit | Required for reference requests | max 16; Reference images as multipart files. Repeat image for multiple files. | Unspecified |
Sizes and aspect ratios
Choose size below. Resolution tiers and aspect ratios are size selectors, not separate request fields.
| Resolution tier | Aspect ratio | size |
|---|---|---|
4K | 1:1 | 4096x4096 |
4K | 3:2 | 4096x2736 |
4K | 2:3 | 2736x4096 |
4K | 4:3 | 4096x3072 |
4K | 3:4 | 3072x4096 |
4K | 16:9 | 4096x2304 |
4K | 9:16 | 2304x4096 |
Results
- Read every item in
data[]. Decode b64_json or download url when present. - This guide documents only the fields and directives listed for each model; it does not define support for other fields such as mask, stream, or tools.
gpt-image-2.5-sunburst Adobe
| Field | Type | Operation | Required | Values and constraints | Recommended |
|---|---|---|---|---|---|
model | string | Generation / reference | Yes | gpt-image-2.5-sunburst Adobe; Exact public model ID. | Unspecified |
prompt | string | Generation / reference | Yes | Describe the image or requested edit. | Unspecified |
n | integer | Generation / reference | No | 1, 2, 3, 4, 5, 6, 7, 8, 9, 10; Requested image count; additional images may increase the charge. | 1 |
size | string | Generation / reference | No | 1024x1024, 1536x1024, 1024x1536, 1152x864, 864x1152, 1536x864, 864x1536, 2048x2048, 2496x1664, 1664x2496, 2304x1728, 1728x2304, 2048x1152, 1152x2048, 4096x4096, 4096x2736, 2736x4096, 4096x3072, 3072x4096, 4096x2304, 2304x4096; Use a listed WIDTHxHEIGHT value. For automatic sizing, omit size; do not send the literal auto. | 1536x1024 |
quality | string | Generation / reference | No | low, medium, high; Use only the values listed for this exact model. | high |
output_format | string | Generation / reference | No | png, jpeg, webp; Encoded image format; separate from the response container. | png |
response_format | string | Generation / reference | No | b64_json; Read every item in data[]. Decode b64_json or download url when present. | b64_json |
image | file[] | edit | Required for reference requests | max 16; Reference images as multipart files. Repeat image for multiple files. | Unspecified |
Results
- Read every item in
data[]. Decode b64_json or download url when present. - This guide documents only the fields and directives listed for each model; it does not define support for other fields such as mask, stream, or tools.
gpt-image-2.5-sunburst Max
| Field | Type | Operation | Required | Values and constraints | Recommended |
|---|---|---|---|---|---|
model | string | Generation / reference | Yes | gpt-image-2.5-sunburst Max; Exact public model ID. | Unspecified |
prompt | string | Generation / reference | Yes | Describe the image or requested edit. | Unspecified |
n | integer | Generation / reference | No | 1, 2, 3, 4, 5, 6, 7, 8, 9, 10; Requested image count; additional images may increase the charge. | 1 |
size | string | Generation / reference | No | 1024x1024, 1536x1024, 1024x1536, 1152x864, 864x1152, 1536x864, 864x1536, 2048x2048, 2496x1664, 1664x2496, 2304x1728, 1728x2304, 2048x1152, 1152x2048, 4096x4096, 4096x2736, 2736x4096, 4096x3072, 3072x4096, 4096x2304, 2304x4096; Use a listed WIDTHxHEIGHT value. For automatic sizing, omit size; do not send the literal auto. | 4096x2736 |
quality | string | Generation / reference | No | auto, low, medium, high, xhigh, max; Use only the values listed for this exact model. | high |
output_format | string | Generation / reference | No | png, jpeg, webp; Encoded image format; separate from the response container. | png |
response_format | string | Generation / reference | No | b64_json; Read every item in data[]. Decode b64_json or download url when present. | b64_json |
image | file[] | edit | Required for reference requests | max 16; Reference images as multipart files. Repeat image for multiple files. | Unspecified |
Results
- Read every item in
data[]. Decode b64_json or download url when present. - This guide documents only the fields and directives listed for each model; it does not define support for other fields such as mask, stream, or tools.
grok-imagine-image-2.0
| Field | Type | Operation | Required | Values and constraints | Recommended |
|---|---|---|---|---|---|
model | string | Generation / reference | Yes | grok-imagine-image-2.0; Exact public model ID. | Unspecified |
prompt | string | Generation / reference | Yes | Describe the image or requested edit. | Unspecified |
n | integer | Generation / reference | No | 1, 2, 3, 4, 5, 6, 7, 8, 9, 10; Requested image count; additional images may increase the charge. | 1 |
size | string | Generation / reference | No | 1024x1024, 1280x720, 720x1280, 1024x768, 768x1024, 1152x768, 768x1152, 2048x2048, 2816x1584, 1584x2816, 2368x1776, 1776x2368, 2496x1664, 1664x2496; Use a listed WIDTHxHEIGHT value. For automatic sizing, omit size; do not send the literal auto. | 2048x2048 |
quality | string | Generation / reference | No | low, medium, auto; Use only the values listed for this exact model. | auto |
output_format | string | Generation / reference | No | png; Encoded image format; separate from the response container. | png |
response_format | string | Generation / reference | No | b64_json; Read every item in data[]. Decode b64_json or download url when present. | b64_json |
image | file[] | edit | Required for reference requests | max 5; Reference images as multipart files. Repeat image for multiple files. | Unspecified |
Sizes and aspect ratios
Choose size below. Resolution tiers and aspect ratios are size selectors, not separate request fields.
| Resolution tier | Aspect ratio | size |
|---|---|---|
1K | 1:1 | 1024x1024 |
1K | 16:9 | 1280x720 |
1K | 9:16 | 720x1280 |
1K | 4:3 | 1024x768 |
1K | 3:4 | 768x1024 |
1K | 3:2 | 1152x768 |
1K | 2:3 | 768x1152 |
2K | 1:1 | 2048x2048 |
2K | 16:9 | 2816x1584 |
2K | 9:16 | 1584x2816 |
2K | 4:3 | 2368x1776 |
2K | 3:4 | 1776x2368 |
2K | 3:2 | 2496x1664 |
2K | 2:3 | 1664x2496 |
Results
- Read every item in
data[]. Decode b64_json or download url when present. - This guide documents only the fields and directives listed for each model; it does not define support for other fields such as mask, stream, or tools.
grok-imagine-nsfw-2.0
| Field | Type | Operation | Required | Values and constraints | Recommended |
|---|---|---|---|---|---|
model | string | Generation / reference | Yes | grok-imagine-nsfw-2.0; Exact public model ID. | Unspecified |
prompt | string | Generation / reference | Yes | Describe the image or requested edit. | Unspecified |
n | integer | Generation / reference | No | 1, 2, 3, 4, 5, 6, 7, 8, 9, 10; Requested image count; additional images may increase the charge. | 1 |
size | string | Generation / reference | No | 1024x1024, 1280x720, 720x1280, 1024x768, 768x1024, 1152x768, 768x1152, 2048x2048, 2816x1584, 1584x2816, 2368x1776, 1776x2368, 2496x1664, 1664x2496; Use a listed WIDTHxHEIGHT value. For automatic sizing, omit size; do not send the literal auto. | 2048x2048 |
quality | string | Generation / reference | No | low, medium, auto; Use only the values listed for this exact model. | auto |
output_format | string | Generation / reference | No | png; Encoded image format; separate from the response container. | png |
response_format | string | Generation / reference | No | b64_json; Read every item in data[]. Decode b64_json or download url when present. | b64_json |
Sizes and aspect ratios
Choose size below. Resolution tiers and aspect ratios are size selectors, not separate request fields.
| Resolution tier | Aspect ratio | size |
|---|---|---|
1K | 1:1 | 1024x1024 |
1K | 16:9 | 1280x720 |
1K | 9:16 | 720x1280 |
1K | 4:3 | 1024x768 |
1K | 3:4 | 768x1024 |
1K | 3:2 | 1152x768 |
1K | 2:3 | 768x1152 |
2K | 1:1 | 2048x2048 |
2K | 16:9 | 2816x1584 |
2K | 9:16 | 1584x2816 |
2K | 4:3 | 2368x1776 |
2K | 3:4 | 1776x2368 |
2K | 3:2 | 2496x1664 |
2K | 2:3 | 1664x2496 |
Results
- Read every item in
data[]. Decode b64_json or download url when present. - This guide documents only the fields and directives listed for each model; it does not define support for other fields such as mask, stream, or tools.
- Reference-image input is not supported.
midjourney
| Field | Type | Operation | Required | Values and constraints | Recommended |
|---|---|---|---|---|---|
model | string | Generation / reference | Yes | midjourney; Exact public model ID. | Unspecified |
prompt | string | Generation / reference | Yes | Describe the image to generate. | Unspecified |
n | integer | Generation / reference | No | ≥ 1; ≤ 50; For one standard Midjourney generation, use n=1; the task returns four candidate images. The current published range is 1–50. | 1 |
mode | string | Generation / reference | No | text-to-image; Generation mode; use text-to-image. This is not a speed setting. | text-to-image |
size | string | Generation / reference | No | Use a supported aspect-ratio string or listed WIDTHxHEIGHT value. Keep size, aspect_ratio, and resolution consistent; final pixel dimensions may differ. | Unspecified |
quality | string | Generation / reference | No | 0.25, 0.5, 1, 2; Quality values are 0.25, 0.5, 1, and 2; default 1. Quality is independent of resolution. | 1 |
resolution | string | Generation / reference | No | 1k, 2k; Use 1k for v6, v6.1, and v7. The 2k option is published for v8.1 and v8.2. | Unspecified |
aspect_ratio | string | Generation / reference | No | 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 4:5, 5:4, 21:9; Supported values: 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 4:5, 5:4, and 21:9. | Unspecified |
style | string | Generation / reference | No | ``, raw; MJ-specific control; its availability depends on the selected version. | Unspecified |
stylize | integer | Generation / reference | No | ≥ 0; ≤ 1000; MJ-specific control; its availability depends on the selected version. | 100 |
chaos | integer | Generation / reference | No | ≥ 0; ≤ 100; MJ-specific control; its availability depends on the selected version. | 0 |
seed | integer | Generation / reference | No | ≥ 0; ≤ 4294967295; MJ-specific control; its availability depends on the selected version. | Unspecified |
sref | string | Generation / reference | No | MJ-specific control; its availability depends on the selected version. | Unspecified |
sw | integer | Generation / reference | No | ≥ 0; ≤ 1000; MJ-specific control; its availability depends on the selected version. | 100 |
iw | number | Generation / reference | No | 0.5, 1, 2, 3; MJ-specific control; its availability depends on the selected version. | 2 |
tile | boolean | Generation / reference | No | MJ-specific control; its availability depends on the selected version. | false |
weird | integer | Generation / reference | No | ≥ 0; ≤ 3000; MJ-specific control; its availability depends on the selected version. | 0 |
negative_prompt | string | Generation / reference | No | MJ-specific control; its availability depends on the selected version. | Unspecified |
hd | boolean | Generation / reference | No | false, true; MJ-specific control; its availability depends on the selected version. | Unspecified |
draft | boolean | Generation / reference | No | false, true; MJ-specific control; its availability depends on the selected version. | Unspecified |
cref | string | Generation / reference | No | MJ-specific control; its availability depends on the selected version. | Unspecified |
cw | integer | Generation / reference | No | 0, 50, 100; ≥ 0; ≤ 100; MJ-specific control; its availability depends on the selected version. | Unspecified |
stop | integer | Generation / reference | No | 50, 80, 100; ≥ 10; ≤ 100; MJ-specific control; its availability depends on the selected version. | Unspecified |
bot_type | string | Generation / reference | No | MID_JOURNEY, NIJI_JOURNEY; MJ-specific control; its availability depends on the selected version. | MID_JOURNEY |
extra | string | Generation / reference | No | Additional MJ command-line options not represented by structured fields. Do not duplicate or conflict with structured parameters. | Unspecified |
count | integer | Generation / reference | No | ≥ 1; ≤ 50; count is an alias for n, with a range of 1–50. Use count=1 for one task, which returns four candidate images. | Unspecified |
output_format | string | Generation / reference | No | png, jpeg; Encoded image format; separate from the response container. | jpeg |
response_format | string | Generation / reference | No | url, b64_json; Read every item in data[]. Decode b64_json or download url when present. | Unspecified |
image | string | string[] | generation | No | max 1; For reference input, send either image or images, not both; the total limit is one image. | Unspecified |
images | ImageInputReference[] | generation | No | max 1; For reference input, send either image or images, not both; the total limit is one image. | Unspecified |
Request example
{
"model": "YOUR_MODEL",
"prompt": "A glass lighthouse above misty cliffs at dawn",
"n": 1,
"size": "16:9",
"aspect_ratio": "16:9",
"resolution": "1k",
"quality": "1",
"style": "raw",
"stylize": 100,
"chaos": 0,
"output_format": "jpeg",
"response_format": "url"
}
Results
- Send a reference as a public HTTPS URL or image Data URL in image or images. The current published model capability lists at most one reference image; a reference guides generation and does not guarantee an exact copy or edit.
- Use structured JSON fields such as size, aspect_ratio, resolution, and quality instead of putting --ar, --v, or --quality in prompt. The selected model determines the version. Use extra for additional MJ command-line options not represented by a structured field.
- With n=1, one Midjourney task returns four images. After completion, process every item in data[] and read url or b64_json according to response_format. No fixed result-URL expiry is stated here; save results you need to keep.
- Controls: style selects raw rendering; stylize controls stylization; chaos controls variation; seed fixes the random seed; sref and sw set a style reference and its weight; iw sets image-reference weight; tile enables seamless tiling; weird controls unusualness; negative_prompt excludes content. Use only the controls published for the selected version.
- Version-specific options: hd is for v8.1/v8.2; draft is for v7/v8.1; cref, cw, and stop are for v6/v6.1; NIJI_JOURNEY is available in bot_type for v6/v7. The version follows the selected model ID; omit version.
- Choose the exact Midjourney model ID exposed to your API key. The version is selected by model; do not use a prompt directive or version field to switch it.
- This guide documents only the fields and directives listed for each model; it does not define support for other fields such as mask, stream, or tools.
nano-banana-2
| Field | Type | Operation | Required | Values and constraints | Recommended |
|---|---|---|---|---|---|
model | string | Generation / reference | Yes | nano-banana-2; Exact public model ID. | Unspecified |
prompt | string | Generation / reference | Yes | Describe the image or requested edit. | Unspecified |
n | integer | Generation / reference | No | 1; Requested image count; additional images may increase the charge. | 1 |
size | string | Generation / reference | No | 1024x1024, 1376x768, 768x1376, 1200x896, 896x1200, 1264x848, 848x1264, 1152x928, 928x1152, 1584x672, 2928x352, 2064x512, 512x2064, 352x2928, 2048x2048, 2752x1536, 1536x2752, 2400x1792, 1792x2400, 2528x1696, 1696x2528, 2304x1856, 1856x2304, 3168x1344, 5856x704, 4128x1024, 1024x4128, 704x5856, 4096x4096, 5504x3072, 3072x5504, 4800x3584, 3584x4800, 5056x3392, 3392x5056, 4608x3712, 3712x4608, 6336x2688, 11712x1408, 8256x2048, 2048x8256, 1408x11712; Use a listed WIDTHxHEIGHT value. For automatic sizing, omit size; do not send the literal auto. | 1024x1024 |
quality | string | Generation / reference | No | low, medium, high; Use only the values listed for this exact model. | medium |
resolution | string | Generation / reference | No | 1K, 2K, 4K; When selecting a listed size, send resolution and aspect_ratio from the same row in the size table. | Unspecified |
aspect_ratio | string | Generation / reference | No | 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 5:4, 4:5, 21:9, 8:1, 4:1, 1:4, 1:8; Use the aspect_ratio from the same row as size and resolution; omit both auxiliary fields when size is omitted. | Unspecified |
output_format | string | Generation / reference | No | png; Encoded image format; separate from the response container. | png |
response_format | string | Generation / reference | No | b64_json; Read every item in data[]. Decode b64_json or download url when present. | b64_json |
image | file[] | edit | Required for reference requests | max 14; Reference images as multipart files. Repeat image for multiple files. | Unspecified |
Sizes and aspect ratios
Use size, resolution and aspect_ratio from the same row. For automatic sizing, omit all three.
| Resolution tier | Aspect ratio | size |
|---|---|---|
1K | 1:1 | 1024x1024 |
1K | 16:9 | 1376x768 |
1K | 9:16 | 768x1376 |
1K | 4:3 | 1200x896 |
1K | 3:4 | 896x1200 |
1K | 3:2 | 1264x848 |
1K | 2:3 | 848x1264 |
1K | 5:4 | 1152x928 |
1K | 4:5 | 928x1152 |
1K | 21:9 | 1584x672 |
1K | 8:1 | 2928x352 |
1K | 4:1 | 2064x512 |
1K | 1:4 | 512x2064 |
1K | 1:8 | 352x2928 |
2K | 1:1 | 2048x2048 |
2K | 16:9 | 2752x1536 |
2K | 9:16 | 1536x2752 |
2K | 4:3 | 2400x1792 |
2K | 3:4 | 1792x2400 |
2K | 3:2 | 2528x1696 |
2K | 2:3 | 1696x2528 |
2K | 5:4 | 2304x1856 |
2K | 4:5 | 1856x2304 |
2K | 21:9 | 3168x1344 |
2K | 8:1 | 5856x704 |
2K | 4:1 | 4128x1024 |
2K | 1:4 | 1024x4128 |
2K | 1:8 | 704x5856 |
4K | 1:1 | 4096x4096 |
4K | 16:9 | 5504x3072 |
4K | 9:16 | 3072x5504 |
4K | 4:3 | 4800x3584 |
4K | 3:4 | 3584x4800 |
4K | 3:2 | 5056x3392 |
4K | 2:3 | 3392x5056 |
4K | 5:4 | 4608x3712 |
4K | 4:5 | 3712x4608 |
4K | 21:9 | 6336x2688 |
4K | 8:1 | 11712x1408 |
4K | 4:1 | 8256x2048 |
4K | 1:4 | 2048x8256 |
4K | 1:8 | 1408x11712 |
Results
- Read every item in
data[]. Decode b64_json or download url when present. - This guide documents only the fields and directives listed for each model; it does not define support for other fields such as mask, stream, or tools.
nano-banana-2.1
| Field | Type | Operation | Required | Values and constraints | Recommended |
|---|---|---|---|---|---|
model | string | Generation / reference | Yes | nano-banana-2.1; Exact public model ID. | Unspecified |
prompt | string | Generation / reference | Yes | Describe the image or requested edit. | Unspecified |
n | integer | Generation / reference | No | 1; Requested image count; additional images may increase the charge. | 1 |
size | string | Generation / reference | No | 1024x1024, 1376x768, 768x1376, 1200x896, 896x1200, 1264x848, 848x1264, 1152x928, 928x1152, 1584x672, 2928x352, 2064x512, 512x2064, 352x2928, 2048x2048, 2752x1536, 1536x2752, 2400x1792, 1792x2400, 2528x1696, 1696x2528, 2304x1856, 1856x2304, 3168x1344, 5856x704, 4128x1024, 1024x4128, 704x5856, 4096x4096, 5504x3072, 3072x5504, 4800x3584, 3584x4800, 5056x3392, 3392x5056, 4608x3712, 3712x4608, 6336x2688, 11712x1408, 8256x2048, 2048x8256, 1408x11712; Use a listed WIDTHxHEIGHT value. For automatic sizing, omit size; do not send the literal auto. | 4800x3584 |
quality | string | Generation / reference | No | low, medium, high; Use only the values listed for this exact model. | high |
resolution | string | Generation / reference | No | 1K, 2K, 4K; When selecting a listed size, send resolution and aspect_ratio from the same row in the size table. | Unspecified |
aspect_ratio | string | Generation / reference | No | 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 5:4, 4:5, 21:9, 8:1, 4:1, 1:4, 1:8; Use the aspect_ratio from the same row as size and resolution; omit both auxiliary fields when size is omitted. | Unspecified |
output_format | string | Generation / reference | No | png; Encoded image format; separate from the response container. | png |
response_format | string | Generation / reference | No | b64_json; Read every item in data[]. Decode b64_json or download url when present. | b64_json |
image | file[] | edit | Required for reference requests | max 14; Reference images as multipart files. Repeat image for multiple files. | Unspecified |
Sizes and aspect ratios
Use size, resolution and aspect_ratio from the same row. For automatic sizing, omit all three.
| Resolution tier | Aspect ratio | size |
|---|---|---|
1K | 1:1 | 1024x1024 |
1K | 16:9 | 1376x768 |
1K | 9:16 | 768x1376 |
1K | 4:3 | 1200x896 |
1K | 3:4 | 896x1200 |
1K | 3:2 | 1264x848 |
1K | 2:3 | 848x1264 |
1K | 5:4 | 1152x928 |
1K | 4:5 | 928x1152 |
1K | 21:9 | 1584x672 |
1K | 8:1 | 2928x352 |
1K | 4:1 | 2064x512 |
1K | 1:4 | 512x2064 |
1K | 1:8 | 352x2928 |
2K | 1:1 | 2048x2048 |
2K | 16:9 | 2752x1536 |
2K | 9:16 | 1536x2752 |
2K | 4:3 | 2400x1792 |
2K | 3:4 | 1792x2400 |
2K | 3:2 | 2528x1696 |
2K | 2:3 | 1696x2528 |
2K | 5:4 | 2304x1856 |
2K | 4:5 | 1856x2304 |
2K | 21:9 | 3168x1344 |
2K | 8:1 | 5856x704 |
2K | 4:1 | 4128x1024 |
2K | 1:4 | 1024x4128 |
2K | 1:8 | 704x5856 |
4K | 1:1 | 4096x4096 |
4K | 16:9 | 5504x3072 |
4K | 9:16 | 3072x5504 |
4K | 4:3 | 4800x3584 |
4K | 3:4 | 3584x4800 |
4K | 3:2 | 5056x3392 |
4K | 2:3 | 3392x5056 |
4K | 5:4 | 4608x3712 |
4K | 4:5 | 3712x4608 |
4K | 21:9 | 6336x2688 |
4K | 8:1 | 11712x1408 |
4K | 4:1 | 8256x2048 |
4K | 1:4 | 2048x8256 |
4K | 1:8 | 1408x11712 |
Results
- Read every item in
data[]. Decode b64_json or download url when present. - This guide documents only the fields and directives listed for each model; it does not define support for other fields such as mask, stream, or tools.
nano-banana-pro
| Field | Type | Operation | Required | Values and constraints | Recommended |
|---|---|---|---|---|---|
model | string | Generation / reference | Yes | nano-banana-pro; Exact public model ID. | Unspecified |
prompt | string | Generation / reference | Yes | Describe the image or requested edit. | Unspecified |
n | integer | Generation / reference | No | 1; Requested image count; additional images may increase the charge. | 1 |
size | string | Generation / reference | No | 1024x1024, 1376x768, 768x1376, 1200x896, 896x1200, 1264x848, 848x1264, 1152x928, 928x1152, 1584x672, 2048x2048, 2752x1536, 1536x2752, 2400x1792, 1792x2400, 2528x1696, 1696x2528, 2304x1856, 1856x2304, 3168x1344, 4096x4096, 5504x3072, 3072x5504, 4800x3584, 3584x4800, 5056x3392, 3392x5056, 4608x3712, 3712x4608, 6336x2688; Use a listed WIDTHxHEIGHT value. For automatic sizing, omit size; do not send the literal auto. | 1024x1024 |
quality | string | Generation / reference | No | low, medium, high; Use only the values listed for this exact model. | medium |
resolution | string | Generation / reference | No | 1K, 2K, 4K; When selecting a listed size, send resolution and aspect_ratio from the same row in the size table. | Unspecified |
aspect_ratio | string | Generation / reference | No | 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 5:4, 4:5, 21:9; Use the aspect_ratio from the same row as size and resolution; omit both auxiliary fields when size is omitted. | Unspecified |
output_format | string | Generation / reference | No | png; Encoded image format; separate from the response container. | png |
response_format | string | Generation / reference | No | b64_json; Read every item in data[]. Decode b64_json or download url when present. | b64_json |
image | file[] | edit | Required for reference requests | max 14; Reference images as multipart files. Repeat image for multiple files. | Unspecified |
Sizes and aspect ratios
Use size, resolution and aspect_ratio from the same row. For automatic sizing, omit all three.
| Resolution tier | Aspect ratio | size |
|---|---|---|
1K | 1:1 | 1024x1024 |
1K | 16:9 | 1376x768 |
1K | 9:16 | 768x1376 |
1K | 4:3 | 1200x896 |
1K | 3:4 | 896x1200 |
1K | 3:2 | 1264x848 |
1K | 2:3 | 848x1264 |
1K | 5:4 | 1152x928 |
1K | 4:5 | 928x1152 |
1K | 21:9 | 1584x672 |
2K | 1:1 | 2048x2048 |
2K | 16:9 | 2752x1536 |
2K | 9:16 | 1536x2752 |
2K | 4:3 | 2400x1792 |
2K | 3:4 | 1792x2400 |
2K | 3:2 | 2528x1696 |
2K | 2:3 | 1696x2528 |
2K | 5:4 | 2304x1856 |
2K | 4:5 | 1856x2304 |
2K | 21:9 | 3168x1344 |
4K | 1:1 | 4096x4096 |
4K | 16:9 | 5504x3072 |
4K | 9:16 | 3072x5504 |
4K | 4:3 | 4800x3584 |
4K | 3:4 | 3584x4800 |
4K | 3:2 | 5056x3392 |
4K | 2:3 | 3392x5056 |
4K | 5:4 | 4608x3712 |
4K | 4:5 | 3712x4608 |
4K | 21:9 | 6336x2688 |
Results
- Read every item in
data[]. Decode b64_json or download url when present. - This guide documents only the fields and directives listed for each model; it does not define support for other fields such as mask, stream, or tools.
seedream-5-0
| Field | Type | Operation | Required | Values and constraints | Recommended |
|---|---|---|---|---|---|
model | string | Generation / reference | Yes | seedream-5-0; Exact public model ID. | Unspecified |
prompt | string | Generation / reference | Yes | Describe the image or requested edit. | Unspecified |
n | integer | Generation / reference | No | 1; Requested image count; additional images may increase the charge. | 1 |
size | string | Generation / reference | No | 2048x2048, 2304x1728, 1728x2304, 2848x1600, 1600x2848, 2496x1664, 1664x2496, 3136x1344, 3072x3072, 3456x2592, 2592x3456, 4096x2304, 2304x4096, 3744x2496, 2496x3744, 4704x2016, 4096x4096, 4704x3520, 3520x4704, 5504x3040, 3040x5504, 4992x3328, 3328x4992, 6240x2656; Use the listed WIDTHxHEIGHT values. The image ratio is encoded in size. | 2048x2048 |
output_format | string | Generation / reference | No | jpeg, png; Encoded image format; separate from the response container. | jpeg |
response_format | string | Generation / reference | No | b64_json; Read every item in data[]. Decode b64_json or download url when present. | b64_json |
image | string | string[] | generation | Required for reference requests | max 14; Reference images as public URL or complete image Data URL strings. | Unspecified |
Sizes and aspect ratios
Choose size below. Resolution tiers and aspect ratios are size selectors, not separate request fields.
| Resolution tier | Aspect ratio | size |
|---|---|---|
2K | 1:1 | 2048x2048 |
2K | 4:3 | 2304x1728 |
2K | 3:4 | 1728x2304 |
2K | 16:9 | 2848x1600 |
2K | 9:16 | 1600x2848 |
2K | 3:2 | 2496x1664 |
2K | 2:3 | 1664x2496 |
2K | 21:9 | 3136x1344 |
3K | 1:1 | 3072x3072 |
3K | 4:3 | 3456x2592 |
3K | 3:4 | 2592x3456 |
3K | 16:9 | 4096x2304 |
3K | 9:16 | 2304x4096 |
3K | 3:2 | 3744x2496 |
3K | 2:3 | 2496x3744 |
3K | 21:9 | 4704x2016 |
4K | 1:1 | 4096x4096 |
4K | 4:3 | 4704x3520 |
4K | 3:4 | 3520x4704 |
4K | 16:9 | 5504x3040 |
4K | 9:16 | 3040x5504 |
4K | 3:2 | 4992x3328 |
4K | 2:3 | 3328x4992 |
4K | 21:9 | 6240x2656 |
Results
- Read every item in
data[]. Decode b64_json or download url when present. - This guide documents only the fields and directives listed for each model; it does not define support for other fields such as mask, stream, or tools.
seedream-v5-pro
| Field | Type | Operation | Required | Values and constraints | Recommended |
|---|---|---|---|---|---|
model | string | Generation / reference | Yes | seedream-v5-pro; Exact public model ID. | Unspecified |
prompt | string | Generation / reference | Yes | Describe the image or requested edit. | Unspecified |
n | integer | Generation / reference | No | 1; Requested image count; additional images may increase the charge. | 1 |
size | string | Generation / reference | No | 1024x1024, 1152x864, 864x1152, 1424x800, 800x1424, 1248x832, 832x1248, 1568x672, 1536x1536, 1792x1344, 1344x1792, 2048x1152, 1152x2048, 1872x1248, 1248x1872, 2352x1008, 2048x2048, 2368x1776, 1776x2368, 2816x1584, 1584x2816, 2496x1664, 1664x2496, 3136x1344; Use a listed WIDTHxHEIGHT value. For automatic sizing, omit size; do not send the literal auto. | 2048x2048 |
output_format | string | Generation / reference | No | png, jpeg; Encoded image format; separate from the response container. | png |
response_format | string | Generation / reference | No | b64_json; Read every item in data[]. Decode b64_json or download url when present. | b64_json |
image | string | string[] | generation | Required for reference requests | max 10; Reference images as public URL or complete image Data URL strings. | Unspecified |
Sizes and aspect ratios
Choose size below. Resolution tiers and aspect ratios are size selectors, not separate request fields.
| Resolution tier | Aspect ratio | size |
|---|---|---|
1K | 1:1 | 1024x1024 |
1K | 4:3 | 1152x864 |
1K | 3:4 | 864x1152 |
1K | 16:9 | 1424x800 |
1K | 9:16 | 800x1424 |
1K | 3:2 | 1248x832 |
1K | 2:3 | 832x1248 |
1K | 21:9 | 1568x672 |
1.5K | 1:1 | 1536x1536 |
1.5K | 4:3 | 1792x1344 |
1.5K | 3:4 | 1344x1792 |
1.5K | 16:9 | 2048x1152 |
1.5K | 9:16 | 1152x2048 |
1.5K | 3:2 | 1872x1248 |
1.5K | 2:3 | 1248x1872 |
1.5K | 21:9 | 2352x1008 |
2K | 1:1 | 2048x2048 |
2K | 4:3 | 2368x1776 |
2K | 3:4 | 1776x2368 |
2K | 16:9 | 2816x1584 |
2K | 9:16 | 1584x2816 |
2K | 3:2 | 2496x1664 |
2K | 2:3 | 1664x2496 |
2K | 21:9 | 3136x1344 |
Results
- Read every item in
data[]. Decode b64_json or download url when present. - This guide documents only the fields and directives listed for each model; it does not define support for other fields such as mask, stream, or tools.
seedream-v5-pro-nsfw
| Field | Type | Operation | Required | Values and constraints | Recommended |
|---|---|---|---|---|---|
model | string | Generation / reference | Yes | seedream-v5-pro-nsfw; Exact public model ID. | Unspecified |
prompt | string | Generation / reference | Yes | Describe the image or requested edit. | Unspecified |
n | integer | Generation / reference | No | 1; Requested image count; additional images may increase the charge. | 1 |
size | string | Generation / reference | No | 1024x1024, 1152x864, 864x1152, 1424x800, 800x1424, 1248x832, 832x1248, 1568x672, 1536x1536, 1792x1344, 1344x1792, 2048x1152, 1152x2048, 1872x1248, 1248x1872, 2352x1008, 2048x2048, 2368x1776, 1776x2368, 2816x1584, 1584x2816, 2496x1664, 1664x2496, 3136x1344; Use a listed WIDTHxHEIGHT value. For automatic sizing, omit size; do not send the literal auto. | 2048x2048 |
output_format | string | Generation / reference | No | png, jpeg; Encoded image format; separate from the response container. | png |
response_format | string | Generation / reference | No | b64_json; Read every item in data[]. Decode b64_json or download url when present. | b64_json |
image | file[] | edit | Required for reference requests | max 10; Reference images as multipart files. Repeat image for multiple files. | Unspecified |
Sizes and aspect ratios
Choose size below. Resolution tiers and aspect ratios are size selectors, not separate request fields.
| Resolution tier | Aspect ratio | size |
|---|---|---|
1K | 1:1 | 1024x1024 |
1K | 4:3 | 1152x864 |
1K | 3:4 | 864x1152 |
1K | 16:9 | 1424x800 |
1K | 9:16 | 800x1424 |
1K | 3:2 | 1248x832 |
1K | 2:3 | 832x1248 |
1K | 21:9 | 1568x672 |
1.5K | 1:1 | 1536x1536 |
1.5K | 4:3 | 1792x1344 |
1.5K | 3:4 | 1344x1792 |
1.5K | 16:9 | 2048x1152 |
1.5K | 9:16 | 1152x2048 |
1.5K | 3:2 | 1872x1248 |
1.5K | 2:3 | 1248x1872 |
1.5K | 21:9 | 2352x1008 |
2K | 1:1 | 2048x2048 |
2K | 4:3 | 2368x1776 |
2K | 3:4 | 1776x2368 |
2K | 16:9 | 2816x1584 |
2K | 9:16 | 1584x2816 |
2K | 3:2 | 2496x1664 |
2K | 2:3 | 1664x2496 |
2K | 21:9 | 3136x1344 |
Results
- Read every item in
data[]. Decode b64_json or download url when present. - This guide documents only the fields and directives listed for each model; it does not define support for other fields such as mask, stream, or tools.