Chat with

ImageGPT Reference

URL-Based Image
Generation

Generate images by constructing a URL. No SDKs, no API keys in headers—just build a URL and use it as an image source.

Psst, Agents should look here → llms.txt

Getting Started

Quick Start

The simplest way to generate an image.

Base URL Format

https://{project}.imagegpt.host/image?prompt={your prompt}

Example

<img src="https://demo.imagegpt.host/image?prompt=A%20sunset%20over%20mountains" alt="Sunset" />

Project slug: Your project slug is the subdomain. Create a project in the dashboard to get your unique slug.

Reference

Parameters

The parameters for image generation.

Parameter	Type	Default	Description
prompt	string	—	Text description of the image to generateRequired
route	string	quality/fast	Predefined route with provider fallback chain. See available routes
model	string	—	Model alias (e.g., flux-2-dev) with automatic provider failover. See available aliases
aspect_ratio	string	1:1	Image aspect ratio. See supported ratios
seed	integer	—	Reproducible generation—same seed produces same image
cache	string	true	Set to 'false' to bypass cache and force new generation

Routing

Routes

Routes provide automatic fallback between providers. If one provider fails, the next in the chain is tried.

quality/fastFast & Cheap

Prioritizes speed and cost. Best for prototyping and high-volume use.

quality/balancedBalanced

Balanced quality, cost, and speed for production use.

quality/highHigh Quality

Premium quality for hero images and marketing assets.

text/fastFast Text

Quick text generation for prototyping.

text/balancedBalanced Text

Good text accuracy with reasonable speed.

text/highHigh Text

Best readable text in images.

realistic/fastFast Photorealism

Quick photorealistic generation.

realistic/balancedBalanced Photorealism

Good photorealism with reasonable speed.

realistic/highMaximum Photorealism

Maximum photorealism for portraits and realistic scenes.

Example

<img src="https://demo.imagegpt.host/image?prompt=A%20sunset&route=quality/high" alt="Sunset" />

Models

Model Aliases

Request a specific model by its canonical name. The system automatically tries multiple providers for reliability and performance.

Gemini

gemini-3-pro-imagePremium

Google Gemini 3 Pro Image — flagship image generation and editing.

gemini-3.1-flash-imageHigh Quality

Google Gemini 3.1 Flash Image — fast, high-quality generation and editing with pro-level quality.

gemini-3.1-flash-lite-imageFast Value

Google Gemini 3.1 Flash Lite Image — fastest, most cost-effective Gemini model with editing and legible text.

gemini-2.5-flash-imageBalanced

Google Gemini 2.5 Flash Image — fast multimodal image generation.

Flux 2

flux-2-devBalanced

Flux 2 Dev — high quality image generation.

flux-2-proPremium

Flux 2 Pro — premium quality image generation.

flux-2-dev-turboFast

Flux 2 Dev Turbo — fast variant of Flux 2.

flux-2-fastFast

Flux 2 Fast — speed-optimized Flux 2.

flux-2-klein-4bUltra-Fast

Flux 2 Klein 4B — ultra-fast 4 billion parameter model.

flux-2-klein-4b-distilledUltra-Fast

Flux 2 Klein 4B Distilled — fastest Klein variant.

flux-2-klein-9bFast

Flux 2 Klein 9B — higher quality 9 billion parameter model.

Flux 1

flux-1-schnellUltra-Fast

Flux 1 Schnell — fastest Flux model.

flux-1.1-pro-ultraPremium

Flux 1.1 Pro Ultra — premium 4MP raw mode.

Other Models

recraft-v3High Quality

Recraft V3 — versatile style-based generation.

ideogram-v3Best Text

Ideogram V3 — industry-leading text rendering.

qwen-image-2512High Quality

Qwen Image 2512 — excellent realism and multilingual text.

seedream-4.5High Quality

Seedream 4.5 — high quality with strong typography.

glm-imageHigh Quality

GLM Image — excellent text rendering and realism.

grok-imagineBalanced

Grok Imagine — highly aesthetic photorealistic images.

juggernaut-flux-proPremium

Juggernaut Flux Pro — maximum photorealism.

imagineart-1.5Balanced

ImagineArt 1.5 — lifelike realism and text rendering.

imagineart-1.5-proPremium

ImagineArt 1.5 Pro — 4K professional-grade visuals.

Example

<img src="https://demo.imagegpt.host/image?prompt=A%20sunset&model=flux-2-dev" alt="Sunset" />

Automatic failover: Model aliases try multiple providers in order. If one provider fails, the next is tried automatically — same reliability as routes.

Dimensions

Aspect Ratios

Control the shape and size of generated images.

1:1

Square (1024x1024)

16:9

Landscape (1344x768)

9:16

Portrait (768x1344)

4:3

Standard (1152x896)

3:4

Tall (896x1152)

3:2

Photo (1216x832)

Response

Response Format

All requests return an image with HTTP response headers conveying information about that request.

Response Headers

Header	Description
Content-Type	image/webp, image/jpeg, or image/png
X-Generation-Time-Ms	Generation duration in milliseconds
X-Request-Id	Unique request identifier for debugging
IGPT-Credits	Credits consumed by this request

Always an image: Every response returns valid image data, making the URL safe to use directly in <img> tags and image contexts without error handling.

Troubleshooting

Errors

When an error occurs, the response still returns an image (a placeholder) with error details in headers.

Invalid Parameters

An unrecognized parameter value never fails the request. The value is replaced with a documented default and the image is generated normally, so a typo in a URL already published on a page does not turn into a broken image for every visitor.

Parameter	Fallback
aspect_ratio	Nearest ratio the model supports, otherwise the model default (`4:4` becomes `1:1`)
model / route	The default route, `quality/fast`
format	The model’s default output format
Numeric parameters	Clamped to the supported range, or dropped when the value is not a number
image_url	Dropped when malformed or unsupported by the model; the prompt alone is generated

Every substitution is reported in an IGPT-Warning response header — one per parameter — so it can be logged or alerted on:

IGPT-Warning: code=PARAM_COERCED; param=aspect_ratio; requested="4:4"; used="1:1"
IGPT-Warning: code=PARAM_DROPPED; param=seed; requested="latest"

The header is present on every response that had a substitution, including cache hits and responses blocked by a security policy — so a blocked request still tells you its parameters were wrong. Only a request that cannot be served at all, such as one with no prompt, returns an error.

Error Headers

Header	Description
IGPT-Error	`code=<CODE>; message="..."` — the error code and a human-readable description
X-Image-Type	`placeholder` when the image returned is an error placeholder
X-Request-Id	Request identifier for debugging

Error Codes

Code	Meaning
VALIDATION_ERROR	Invalid or missing parameters
PROJECT_NOT_FOUND	Invalid project slug in subdomain
POLICY_BLOCKED	Blocked by a project security policy
LIMIT	Rate limit exceeded
TRIAL_CREDITS_EXHAUSTED	Trial credits used up
CREDIT_LIMIT_REACHED	Monthly credit limit reached
GENERATION_FAILED	All providers failed to generate

Error Image

When an error occurs, an error image is returned instead of a broken image. The image includes the error category and a brief description:

Examples

Full Examples

Basic HTML

<img
  src="https://demo.imagegpt.host/image?prompt=A%20sunset%20over%20mountains"
  alt="Sunset over mountains"
/>

High quality with aspect ratio

<img
  src="https://demo.imagegpt.host/image?prompt=Professional%20headshot&route=quality/high&aspect_ratio=3:4"
  alt="Professional headshot"
/>

Text rendering (sign, label, etc.)

<img
  src="https://demo.imagegpt.host/image?prompt=A%20neon%20sign%20that%20says%20OPEN&route=text/high"
  alt="Neon open sign"
/>

URL-Based ImageGeneration

Quick Start

Base URL Format

Example

Parameters

Routes

Example

Model Aliases

Gemini

Flux 2

Flux 1

Other Models

Example

Aspect Ratios

Response Format

Response Headers

Errors

Invalid Parameters

Error Headers

Error Codes

Error Image

Full Examples

Basic HTML

High quality with aspect ratio

Text rendering (sign, label, etc.)

URL-Based Image
Generation