Hermosso

"Macro bioluminescent bloom"

Generate with Hermosso

Hermosso

"Macro bioluminescent bloom"

Generate with Hermosso

HomeImage Models › GPT Image 2

GPT Image 2 on Hermosso

OpenAI's image model: the strongest instruction-follower in the studio, with text that actually reads.

OpenAI · September 5, 2026

An image generated with GPT Image 2 — a vintage Monaco Grand Prix poster with legible typography

Try GPT Image 2

What GPT Image 2 does best

It follows the brief

GPT Image 2 is the strongest instruction-follower we've tested: complex, multi-part prompts — 'a poster with this exact headline, three objects arranged left to right' — come back right more often than with anything else.

A poster with legible text generated with GPT Image 2

Text and graphics that read

Headlines, labels, layouts — when the image has to contain words, GPT Image 2 is one of the two models we reach for (Ideogram is the other). The Monaco poster above is a single generation, not a composite.

Reference-capable

GPT Image 2 accepts reference images, so it edits as well as it generates: upload a photo and ask for changes, compositions or fusions.

Worth knowing: the pace

Expect around a minute per image at 100 credits — deliberate rather than instant. For drafts and variations, run the fast tier first and bring the winning prompt back here.

Made for the work you ship

  • Posters with real headlines

    Exact wording, arranged where you asked — the brief-following strength shows most on text-heavy work.

  • Complex scenes

    Multi-subject, multi-constraint prompts are where weaker models improvise; GPT Image 2 stays on-script.

  • Structured edits

    Attach a reference and describe precise changes — instruction-following applies to edits too.

Specs at a glance

  • OpenAI
  • Best instruction-following
  • Legible text
  • Reference edits
  • 100 credits / image

How to use it

  1. Open Hermosso's Create studio and write your prompt — be as specific as you like; this model rewards detail.
  2. Attach a reference photo if the job starts from an existing image.
  3. Pick GPT Image 2 and generate — 100 credits per image, around a minute each. New accounts get 750 free credits.

Frequently asked questions

What is GPT Image 2?

GPT Image 2 is OpenAI's image generation model — in our studio tests, the strongest instruction-follower available: complex multi-part prompts and exact text come back right more often than with any other model. It costs 100 credits per image on Hermosso.

GPT Image 2 vs GPT Image 1 — what's new?

GPT Image 2 is OpenAI's current image model — the successor generation with sharper instruction-following and text rendering. It's the version available in Hermosso's Create studio.

How do I use GPT Image 2 on Hermosso?

Open the Create studio, write a prompt (the more specific, the better), optionally attach a reference photo, and pick GPT Image 2 — 100 credits per image. New accounts get 750 free credits, and plans start at $9/month.

Is GPT Image 2 good at text in images?

Yes — it's one of the two models we recommend for text-heavy work (Ideogram is the other, especially for posters and logos). Exact headlines and arranged layouts are its signature strength.

GPT Image 2 vs Nano Banana 2 — which should I pick?

GPT Image 2 for complex prompts, text and exact layouts; Nano Banana 2 for maximum resolution (up to 4K) and likeness-keeping photo edits. Both cost 100 credits at their base tier — test the same prompt on both.

Try GPT Image 2 on Hermosso

One credit balance across every model in the studio. New accounts get 750 free credits — and credits never expire.

Try GPT Image 2