GPT-Image-2 – Artlist

GPT-Image-2 is OpenAI’s next-generation flagship image model that sets a new industry standard for photorealism and near-perfect text rendering.

Key Features

Text-to-Image, Image-to-Image
Conversational Prompt Adherence: Handles natural-language prompts.
Text Rendering: near-perfect text rendering on signs, UI text, and infographics.
Transparent Background: Generate images with transparent or opaque backgrounds on demand.
Image Editing:
Natural Language Editing: Describe changes in plain text; composition and lighting are preserved.
Speed:
4× Faster Generation: Reported 4× speed improvement over prior GPT-Image versions.
Quality: Native output up to 3840px with three quality tiers (low, medium, high)

Inputs: Text-to-Image · Image-to-Image
Resolution:
- 1K Quality - low
  - 1:1 → 1080 x 1080
  - 3:4 → 1080 x 1440
  - 4:3 → 1440 x 1080
  - 16:9 → 1920 x 1080
  - 9:16 → 1080 x 1920
- 2K Quality - medium
  - 1:1 → 2048 x 2048
  - 3:4 → 1728 x 2304
  - 4:3 → 2304 x 1728
  - 16:9 → 2560 x 1440
  - 9:16 → 1440 x 2560
- 4K Quality (UHD) - high
  - 1:1 → 2160 x 2160
  - 3:4 → 2160 x 2880
  - 4:3 → 2880 x 2160
  - 16:9 → 3840 x 2160
  - 9:16 → 2160 x 3840
Aspect Ratios: 1:1, 3:4, 4:3, 16:9, 9:16
Quality: three levels of quality (low, medium and high) with up to 3840px resolution for 4K UHD.

Prompt Formula: [Core subject] + [Style/mood] + [Specific elements] + [Quality level]

Use conversational phrasing — the model handles natural language better than keyword lists.
Quality parameter accepts 'low', 'medium', or 'high'; default is 'high'.
For edits, input_fidelity 'high' preserves more of the source; 'low' gives the model more creative freedom.
Background parameter: set to 'transparent' for compositing layers, 'opaque' to force a solid background.