Skip to main content
The /v1/images/generations endpoint provides OpenAI Images API compatibility to generate images from natural language descriptions.

Request Parameters

Technical Architecture & Caching

  • Synchronous vs Asynchronous: The standard /v1/images/generations endpoint operates synchronously for real-time applications. For high-resolution generation or batch pipelines, use the Responses API with background: true to prevent HTTP timeouts.
  • Ephemeral CDN & TTL Caching: Generated images are delivered through high-speed edge CDN storage via signed URLs. Links remain valid for 5 minutes (TTL). Ingest assets into permanent cloud storage (S3/GCS/R2) if long-term persistence is required.
  • SSRF Mitigation: When providing input image URLs for variations or inpainting, the gateway security bridge validates public IP ranges and blocks loopback or private addresses to prevent Server-Side Request Forgery.
  • Rupiah Metering: Image generation is billed on a per-image unit basis rather than token count. Costs are deducted directly from your prepaid Rupiah balance.

Visual Showcase

Output examples generated directly through the Neosantara API:

Fast Inference

Model: imagen-4.0-fast
Imagen 4.0 Fast output showcase

Prompt: 'A vibrant digital painting of a high-tech laboratory in Jakarta, neon highlights, futuristic design, extremely detailed.'

Artistic & Fantasy

Model: sdxl-lightning
Artistic fantasy output showcase

Prompt: 'A majestic Indonesian temple in the clouds, ethereal lighting, high fantasy art style, 8k.'

Realistic Photography

Model: titan-image-g1
Realistic photography output showcase

Prompt: 'A realistic high-detail photograph of a Balinese traditional dancer performing at night, focused lighting.'

Image-to-Image Transformation

Model: stable-diffusion-v1-5-img2img Transform existing images into new visual styles while preserving original structural composition.
showcase-titan-base

Original Image

Input reference photograph of a Balinese dancer.
showcase-img2img-result

Sketch Transformation

Stylized output transformed into a minimalist charcoal sketch.

Available Image Generation Models