Skip to main content
POST
/api/v1/image/generate

Sizing Options

Image models use model-specific sizing parameters:
  • Pixel-based models accept width and height, for example venice-sd35.
  • Aspect-ratio models accept aspect_ratio, for example qwen-image-2. The Qwen Image family — qwen-image, qwen-image-3, qwen-image-3-pro, qwen-image-2-1-turbo, and qwen-image-2-1-pro — rejects width and height with a 400; use aspect_ratio for those.
  • Resolution-tier models accept both aspect_ratio and resolution.
For example, gpt-image-2, nano-banana-2, and nano-banana-pro support resolution values of 1K, 2K, and 4K:
The same resolution-tier sizing pattern also applies to Nano Banana models:
For models that accept aspect ratio but do not expose resolution tiers, send aspect_ratio without resolution:
For models that use direct pixel dimensions, send width and height instead:
Check Image Models or the Models endpoint for each model’s supported sizing options before mixing sizing fields across models.

Quality Tiers

Some models accept an optional quality parameter (low, medium, high) that trades visual fidelity for cost. Models that list qualities in model_spec.constraints from GET /models?type=image accept only those values, and some support only a subset such as low and medium. Other models ignore the parameter.
When you omit quality, the model uses its default tier (high for gpt-image-2). Pricing depends on the combination of resolution and quality — the per-tier prices for gpt-image-2 (1K/2K/4K × low/medium/high) are exposed under model_spec.pricing.quality in the Models endpoint and listed on the GPT Image 2 page.
Quality-aware pricing is the default for all API and SDK callers — no opt-in header required. Requests that omit quality are billed at the model’s default tier (high for gpt-image-2).

Postman Collection

For additional examples, please see this Postman Collection.

Authorizations

Authorization
string
header
required

Bearer authentication header of the form Bearer <token>, where <token> is your auth token.

Headers

Accept-Encoding
string

Supported compression encodings (gzip, br). Only applied when return_binary is false.

Example:

"gzip, br"

Body

application/json
model
string
required

The model to use for image generation.

Example:

"grok-imagine-image"

prompt
string
required

The description for the image. Character limit is model specific and is listed in the promptCharacterLimit setting in the model list endpoint.

Required string length: 1 - 7500
Example:

"A beautiful sunset over a mountain range"

anon_user_id
string

Optional identifier for the API customer's end user. Combined with the Venice user id when attributing the request to upstream providers. Distinct from the discarded OpenAI user field. Must be printable ASCII and must not contain ||.

Required string length: 1 - 128
Pattern: ^[\x20-\x7E]+$
Example:

"end-user-123"

cfg_scale
number

CFG scale parameter. Higher values lead to more adherence to the prompt.

Required range: 0 < x <= 20
Example:

7.5

embed_exif_metadata
boolean
default:false

Embed prompt generation information into the image's EXIF metadata.

Example:

false

format
enum<string>
default:webp

The image format to return. WebP are smaller and optimized for web use. PNG are higher quality but larger in file size.

Available options:
jpeg,
png,
webp
Example:

"webp"

height
integer
default:1024

Height of the generated image.

Required range: 0 < x <= 1280
Example:

1024

hide_watermark
boolean
default:false

Whether to hide the Venice watermark. Venice may ignore this parameter for certain generated content.

Example:

false

inpaint
any | null
deprecated

This feature is deprecated and was disabled on May 19th, 2025. A revised in-painting API will be launched in the near future.

lora_strength
integer

Lora strength for the model. Only applies if the model uses additional Loras.

Required range: 0 <= x <= 100
Example:

50

negative_prompt
string

A description of what should not be in the image. Character limit is model specific and is listed in the promptCharacterLimit constraint in the model list endpoint.

Maximum string length: 7500
Example:

"Clouds, Rain, Snow"

return_binary
boolean
default:false

Whether to return binary image data instead of base64.

Example:

false

variants
integer

Number of images to generate (1–4). Only supported when return_binary is false. Variants the model provider explicitly rejects under its content policy are omitted from a successful partial response. Other moderation outcomes can still return blacked-out images; check x-venice-is-content-violation.

Required range: 1 <= x <= 4
Example:

3

safe_mode
boolean
default:true

Whether to blur images classified as adult content. Defaults to true. Set to false to receive unblurred output. No Pro subscription is required, and the mature filter setting in the Venice app does not affect this. The x-venice-is-blurred response header reports whether the returned image was blurred.

Example:

false

seed
integer
default:0

Random seed for generation. If not provided, a random seed will be used.

Required range: -999999999 <= x <= 999999999
Example:

123456789

steps
integer
default:8

Number of inference steps. This model does not support steps - this field is ignored.

Example:

8

style_preset
string

An image style to apply to the image. Visit https://docs.venice.ai/api-reference/endpoint/image/styles for more details.

Example:

"3D Model"

aspect_ratio
string

Aspect ratio (utilized by certain image models including Nano Banana). Examples: "1:1", "16:9".

Example:

"1:1"

resolution
string

Resolution (utilized by certain image models including Nano Banana). Examples: "1K", "2K", "4K".

Example:

"1K"

quality
enum<string>

Output quality for supported models (e.g. GPT Image 2 / GPT Image 2 Edit). Higher values can increase the final request charge. See the model list for supported options.

Available options:
low,
medium,
high
Example:

"high"

transparent_background
boolean

Return the image with a transparent background (alpha channel). Only supported by models with supportsTransparentBackground: true in their constraints (see GET /api/v1/models), and requires format png or webp; otherwise the request is rejected. Omit or set to false to keep the model default background.

Example:

false

enable_web_search
boolean

Enable web search for the image generation task. This will allow the model to use the latest information from the web to generate the image. Only supported by certain models. If web search is used, additional credits are getting charged.

Example:

false

disable_prompt_optimization_thinking
boolean

Skip the model's prompt-optimization thinking step for faster generation. Only supported by models with supportsOptimizePromptThinking; ignored by others. Omit to use the model default.

Example:

false

enhance_prompt
boolean

Rewrite the prompt before generation to add clarifying visual detail. Additional credits are charged when a rewrite is generated. Adds up to ~30 seconds before generation starts. When enhancement is applied, the final prompt is returned URL-encoded in the x-venice-enhanced-prompt response header. Defaults to false.

Example:

false

width
integer
default:1024

Width of the generated image.

Required range: 0 < x <= 1280
Example:

1024

style_references
object[]

Style reference images that guide the aesthetic of the generated image. Only supported by models with supportsStyleReferences: true (see GET /api/v1/models for the per-model maxStyleReferences limit).

Example:

Response

Successfully generated image

id
string
required

The ID of the request.

Example:

"generate-image-1234567890"

images
string[]
required

Base64 encoded image data. A partial response can contain fewer entries than requested.

timing
object
required
content_policy_violation_count
integer

Number of requested variants omitted because the model provider explicitly rejected them under its content policy. Other moderation outcomes can still return blacked-out images flagged by x-venice-is-content-violation.

Required range: x >= 1
Example:

1

request
any | null

The original request data sent to the API.