---
title: "Unified Image Generate"
method: POST
path: "/v1/images/generate"
tags: ["Unified Agent APIs"]
---

# Unified Image Generate

`POST /v1/images/generate`

Generate images from text prompts. Returns base64-encoded image bytes.

Response data contains b64_json (base64 string) and content_type (MIME type).
Uses Gemini Flash models only (retried); fails gracefully so callers continue without an image.

## Request body

- ImageGenerateRequest
  - `prompt` string, required — Describe a CONCRETE, grounded scene: real people, real objects, physical screens and devices, natural light, editorial photography quality (or a clean minimal illustration of the actual concept). Words like 'futuristic', 'holographic', 'glowing', or neon-tech scenery produce the same generic AI-cliche image every time; describe the present-day, physical version of the scene instead.
  - `aspect_ratio` string
  - `image_size` string
  - `reference_image_url` string, nullable
  - `skip_default_style` boolean — Leave false (recommended). False applies the house visual style guidance (grounded realistic imagery, avoids overused AI-tech cliches). Set true ONLY when you are deliberately providing complete art direction of your own and want zero house styling.

## Response `200`

Successful Response

- unknown

## Other responses

- `422` — Validation Error

---

[API](https://skmtc.net/suprsonic/apis/suprsonic-unified-agent-api.md) · [All operations](https://skmtc.net/suprsonic/apis/suprsonic-unified-agent-api/llms.txt) · [OpenAPI document](https://skmtc-service-staging.skmtc.workers.dev/v1/apis/suprsonic/suprsonic-unified-agent-api/revisions/2a1b6fcb4ef0/schema)