Supported providers and models

OpenAI

• Base URL: https://api.openai.com/v1
• Generation: Yes
• Editing: Yes

Models

Model	Features
gpt-image-1	Full token usage tracking, editing support
dall-e-3	revised_prompt in response
dall-e-2	Legacy model

Generation parameters

Parameter	Type	Default	Description
model	string	gpt-image-1	Model name
n	integer	1	Number of images
size	string	auto	Image dimensions
quality	string	auto	Image quality (standard, hd, auto)
style	string	—	Image style (natural, vivid)
response_format	string	—	url or b64_json
background	string	—	Background setting
moderation	string	—	Moderation level
output_compression	integer	—	Output compression level
output_format	string	—	Output format (png, jpeg)

Edition parameters

Parameter	Type	Default	Description
model	string	gpt-image-1	Model name
n	integer	—	Number of results
size	string	—	Image dimensions
quality	string	—	Image quality
background	string	—	Background setting
response_format	string	—	url or b64_json

---

Azure OpenAI

• Base URL: Computed from Azure resource configuration
• Generation: Yes
• Editing: No

Connection

Azure OpenAI requires specific connection settings:

{
  "connection": {
    "resource_name": "my-azure-resource",
    "deployment_id": "my-deployment",
    "api_version": "2024-02-01",
    "api_key": "xxx"
  }
}

Parameter	Type	Default	Description
resource_name	string	—	Azure resource name
deployment_id	string	—	Deployment/model ID
api_version	string	2024-02-01	API version
api_key	string	—	API key (alternative to bearer token)

Generation parameters

Same as OpenAI, with the following constraint:

Allowed sizes: 1024x1024, 1792x1024, 1024x1792 only. Any other value defaults to 1024x1024.

Models

Model	Features
gpt-image-1	Full token usage tracking
dall-e-3	revised_prompt in response
dall-e-2	Legacy model

---

Cloud Temple

• Base URL: https://api.ai.cloud-temple.com/v1
• Generation: Yes
• Editing: Yes

Cloud Temple exposes an OpenAI-compatible API. It uses the same parameters and options as OpenAI.

---

Gemini

• Base URL: https://generativelanguage.googleapis.com/v1beta/openai
• Generation: Yes
• Editing: Yes

Gemini exposes an OpenAI-compatible image API. It uses the same parameters and options as OpenAI.

Models

Model	Description
imagen-3.0-generate-002	Default Imagen 3 model
google/nano-banana-pro	Gemini 3 Pro Image via AI/ML API

---

Grok (X-AI)

• Base URL: https://api.x.ai
• Generation: Yes
• Editing: No

Models

Model	Description
grok-2-image	Default Grok image model

Generation parameters

Grok has a simpler parameter set:

Parameter	Type	Default	Description
model	string	grok-2-image	Model name
n	integer	1	Number of images
response_format	string	url	url or b64_json

Size, quality, and style parameters are not supported.

---

Luma

• Base URL: https://api.lumalabs.ai
• Generation: Yes
• Editing: No

Luma uses the Dream Machine API for image generation.

Models

Model	Description
photon-1	Default Luma model
photon-flash-1	Fast generation model

Generation parameters

Parameter	Type	Default	Description
model	string	photon-1	Model name

Only prompt and model are supported. Size, quality, and other parameters are not available.

---

Hive

• Base URL: https://api.thehive.ai/api/v3
• Generation: Yes
• Editing: No

Hive uses a unique API where the model name is part of the URL path.

Models

Model	Description
black-forest-labs/flux-schnell	Default model (FLUX Schnell)

Generation parameters

Parameter	Type	Default	Description
model	string	black-forest-labs/flux-schnell	Model name (used as URL path segment)
num_images	integer	—	Number of images
image_size.width	integer	1024	Image width in pixels
image_size.height	integer	1024	Image height in pixels
num_inference_steps	integer	—	Number of inference steps
seed	integer	—	Random seed for reproducibility
output_quality	integer	90	Output quality (0-100)
output_format	string	jpeg	Output format

Note: Hive uses width/height integers instead of a size string like 1024x1024.

---

Arkane Cloud

• Base URL: https://console.arkanecloud.com/api/v2
• Generation: Yes
• Editing: No

Arkane Cloud serves open diffusion models (Stable Diffusion, FLUX...) billed per image. Consumers keep using the standard OpenAI images API: the gateway turns the OpenAI size (like 1024x768) into the width / height Arkane Cloud expects, and forwards the diffusion settings (inference steps, guidance, negative prompt, seed) when they are set in the request or in the model configuration.

{
  "connection": {
    "token": "xxx"
  },
  "options": {
    "generation": {
      "enabled": true,
      "model": "stability-ai/sdxl",
      "width": 1024,
      "height": 1024,
      "response_format": "b64_json"
    },
    "edition": { "enabled": false }
  }
}

Models

Model	Description
stability-ai/sdxl	Default model (Stable Diffusion XL)

The full list of image models is available in the Arkane Cloud console.

Generation parameters

Parameter	Type	Default	Description
model	string	stability-ai/sdxl	Model name
size	string	—	Image dimensions like 1024x768, used when width / height are not set
width	integer	1024	Image width in pixels (64 to 2048)
height	integer	1024	Image height in pixels (64 to 2048)
num_inference_steps	integer	model default	Number of inference steps
guidance_scale	number	model default	How closely the image follows the prompt
negative_prompt	string	—	What the image should avoid
seed	integer	—	Random seed for reproducibility (-1 for a random one)
response_format	string	b64_json	b64_json or url
response_extension	string	png	Image file format, also set by the OpenAI output_format of the request

Every parameter can be set in the model configuration and overridden per request. Arkane Cloud generates one image per request, so n, quality and style are not used.

---

OpenAI Compatible

• Base URL: your own endpoint (set in the connection config)
• Generation: Yes
• Editing: Yes

A generic provider for any image service that speaks the OpenAI images API. Point it at your endpoint with base_url and token, and give it a display name with provider_name — that name is what shows up for this provider everywhere in the gateway. When the target API needs it, you can also override the request headers, the param_mappings, and the additional_body_params. It uses the same parameters and options as OpenAI; the actual feature set depends on the target endpoint.

{
  "connection": {
    "base_url": "https://my-endpoint/v1",
    "provider_name": "My Provider",
    "token": "xxx"
  },
  "options": {
    "generation": { "enabled": true, "model": "your-image-model", "n": 1 },
    "edition": { "enabled": false }
  }
}

---

OVH AI Endpoints 🇫🇷 🇪🇺

• Base URL: https://oai.endpoints.kepler.ai.cloud.ovh.net/v1
• Generation: Yes
• Editing: Yes

OVH AI Endpoints exposes a unified OpenAI-compatible image API. It uses the same parameters and options as OpenAI — just set your token and the image model you want to use.

---

Provider comparison

Feature	OpenAI	Azure	Cloud Temple	Gemini	Grok	Luma	Hive	Arkane	OVH	OAI Compat
Generation	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Editing	Yes	No	Yes	Yes	No	No	No	No	Yes	Yes
Size control	Yes	Yes (restricted)	Yes	Yes	No	No	Yes (w/h)	Yes	Yes	Yes
Quality control	Yes	Yes	Yes	Yes	No	No	Yes	No	Yes	Yes
Style control	Yes	Yes	Yes	Yes	No	No	No	No	Yes	Yes
Seed/reproducibility	No	No	No	No	No	No	Yes	Yes	No	No
Inference steps	No	No	No	No	No	No	Yes	Yes	No	No

For OpenAI Compatible, the values above reflect the OpenAI parameters the gateway forwards — what actually works depends on the endpoint you point it at.