OpenAI image models understand layered instructions like a conversation partner: complex scenes with multiple conditions, infographics, precise text labels.
In the studio, open the image model catalog and find GPT Image:
GPT Image 2 parameters:
The model's strength is executing multi-layered instructions without compromise:
Write like a design brief — structured and complete.
Hit "Generate" — images appear in the studio gallery within 5–30 seconds. Any image can be sent with one click to a video model as a starting frame, to Kling Elements as a character, or downloaded at full resolution.
How to generate an image in GPT Image 2 online?
Open Airium Studio, select GPT Image 2 in the image model catalog, write your prompt, and click "Generate". No API keys or VPN required — new users get free tokens.
How much does generation cost in GPT Image 2?
Price depends on the selected quality (low/medium/high) and size — from a couple of tokens for a draft to ~15 for high. You only pay for successful generations.
Does GPT Image 2 support reference images?
GPT Image 2 accepts reference images and can redraw them following instructions.
Can I use GPT Image 2 from Russia without a VPN?
Yes — Airium Studio works in the browser, payment via Russian bank cards and cryptocurrency, no VPN needed.
| Parameter | Values | Tip |
|---|---|---|
| Sizes | 1024² · 1536×1024 · 1024×1536 | square/landscape/portrait |
| Quality | low / medium / high | quality = price |
| Strength | multi-layered briefs | requirements list in the prompt |
| DALL·E 3 | illustrations | stylization and art |
Charts, diagrams, slides — with accurate labels.
Logo concepts, covers with precise text.
Structure: "what + where + style + what NOT to do".
low for idea exploration, high for production.