Skip to main content
gpt image 2gpt-image-2gpt image 2.5gpt image 2.6gpt image 2 promptschatgpt images 2.6openai image modelgpt image flare sunburst

GPT Image 2, 2.5 and 2.6: What Changed, What It Costs, and How to Prompt It

OpenAI's GPT Image 2, GPT Image 2.5 and GPT Image 2.6 explained: release dates, Flare vs Sunburst, resolution, price per image, what each one is best and worst at, and prompts that keep a real face looking like itself.

Starrd Team|November 19, 20265 min read

Quick answer

GPT Image 2 is OpenAI's image model from April 2026. GPT Image 2.5 (September 8, 2026) and GPT Image 2.6 (November 4, 2026) each ship in two versions: Flare, built for speed, and Sunburst, built for quality and precise edits. All three take up to 16 reference images and lead at readable text and at keeping a face during edits. GPT Image 2.6 Sunburst is first on the Artificial Analysis text-to-image leaderboard but uses several times more tokens per image; GPT Image 2.5 costs a flat 3 to 8 cents per image through resellers and is the better value. Prompt it with labeled sections, exact text in quotes, numbered references with a job each, and an explicit list of what must not change.

OpenAI shipped three versions of its image model in seven months. Here's what actually changed between them, what each one costs, and how to prompt them, from a team that renders well over a thousand GPT Image stills a week. Facts as of November 19, 2026.

The three versions

GPT Image 2GPT Image 2.5GPT Image 2.6
ReleasedApril 21, 2026September 8, 2026November 4, 2026
VersionsOneFlare (speed), Sunburst (quality)Flare, Sunburst
Text-to-image rank#6#2 (Sunburst)#1 (Sunburst)
Editing rank#13#4#2
New16 references, strong textHigher quality, up to ~3840 px edges50% faster, GIF and SVG, complexity setting
Price on KIE$0.03 / $0.05 / $0.08$0.03 / $0.05 / $0.08Token-billed

Ranks: Artificial Analysis, read November 19. Prices per image at 1K / 2K / 4K on KIE. Release notes: OpenAI, OpenAI docs.

Which one should you use? GPT Image 2.5 for almost everything. It's within a few points of 2.6 on blind votes at a flat, predictable price. Use 2.6 Sunburst when a single image has to be as good as possible and cost doesn't matter, or when you need GIF or SVG output.

What it's best at

Text you can read. Posters, menus, game screens and infographics. The GPT models are the most reliable at small, correct lettering.

Before — the original photo

Same source photo

GPT Image 2: a selfie turned into a WANTED card

After — GPT Image 2: a selfie turned into a WANTED card
After

The face holds and the card text is clean, the two things GPT Image does better than anyone.

Keeping a real face. When you list what must not change, GPT Image holds identity better than Nano Banana or Seedream. That's why we moved almost every Starrd template's first frame to GPT Image 2.5 on November 17: on the same photos and prompts, Nano Banana Pro returned cartoon-looking character sheets on templates that asked for photoreal, and GPT 2.5 didn't.

Many references. Up to 16 images in one edit: a face, an outfit, a room, a style frame.

What it's bad at

  • Grain. When ChatGPT Images 2.5 launched, Reddit users complained about noisy skies and flat backgrounds, and TechRadar's own tests agreed.
  • Plastic skin in product edits, and faces drifting when several people share the frame (OpenAI forum).
  • Refusals. It turns down some costumes and heavy makeup other models allow. We keep two templates on Nano Banana Pro for exactly that: a zombie makeup look GPT dropped, and costumes it refused.
  • Cost surprises on 2.6. Community tests at 1K found 2.6 uses about 4.2 times the output tokens of 2.5 at low quality and 3.3 times at medium. OpenAI hasn't confirmed this, so check usage.output_tokens on a real call before you budget.
  • 4:5 on 2.5. In our tests GPT Image 2.5 rejected 4:5 even though the docs list it. Use 3:4 for portrait.

How to prompt GPT Image

These come from OpenAI's image prompting guide and from our own templates.

  1. Use labeled sections. Scene, subject, details, constraints. The model follows structure better than one long sentence.
  2. Put exact text in quotes, then say "no other text". Use medium quality or higher for small lettering.
  3. Number your references and give each a job. "Image 1 is the subject. Image 2 is the outfit. Image 3 is the background."
  4. Say what changes, then what doesn't. "Change only the jacket to red leather. Do not change the face, facial features, skin tone, hair, body shape, pose or identity."
  5. One change per edit. Five changes at once is how faces drift.
  6. Ask for photoreal explicitly, with "natural skin texture, visible pores", or GPT tends toward a polished look.
  7. On 2.6, try the complexity setting. It defaults to high. Compare low and high at the same quality before you decide it's worth the tokens.
Edit prompt: keep the person, change the scene
Image 1 is the subject. Keep their face, facial features, skin tone, hair and body shape exactly as in Image 1.Scene: the subject standing on a red carpet at a film premiere, photographers behind a rope, flashes going off.
Details: black tuxedo, slight smile, looking just past the camera. Natural skin texture, visible pores, photoreal.
Text: a step-and-repeat banner behind them reads "PREMIERE" in white. No other text.
Format: vertical 3:4.

Price: what you'll actually pay

  • OpenAI API: $5 per million text input tokens, $8 per million image input tokens, $30 per million image output tokens, the same for 2, 2.5 and 2.6. Batch halves output.
  • Resellers: KIE sells GPT Image 2 and 2.5 at a flat $0.03 / $0.05 / $0.08 per image (1K / 2K / 4K), and 2.6 by token (plus a $0.025 "Economy" 1K tier).

Using GPT Image 2.5 in Starrd

You don't need an API key. Starrd's Custom mode has GPT Image 2.5 as its default photo model: add up to 10 photos, write the scene, and get 2K images for 10 credits for one or 30 for four.

Open the red carpet prompt in Starrd

Every video template works the same way underneath: GPT Image 2.5 draws you into the first frame, then a video model (Seedance, Kling or Grok) animates it.

Red Carpet: GPT Image 2.5 first frame, animated into video

One selfie in. The still is GPT Image 2.5; the motion is the video model. 100 credits.

Make this

Want a recurring character instead of a one-off? Character Studio makes a portrait plus a four-view character sheet with GPT Image 2, for 20 credits.

Compare it with the rest of the field in the best AI image generators of 2026.

Frequently Asked Questions

What is the difference between GPT Image 2, 2.5 and 2.6?

GPT Image 2 (April 21, 2026) is a single model. GPT Image 2.5 (September 8, 2026) split into Flare for speed and Sunburst for quality and edit precision, and ranks higher on blind votes. GPT Image 2.6 (November 4, 2026) keeps the Flare and Sunburst pair, renders about 50% faster, adds animated GIF and SVG output and a complexity setting, and is first on the Artificial Analysis text-to-image leaderboard.

Is GPT Image 2.6 real?

Yes. OpenAI released Images 2.6 on November 4, 2026, as gpt-image-2.6 Flare and Sunburst in the API and as Images 2.6 in ChatGPT.

How much does GPT Image 2 cost?

OpenAI bills all three versions by token at the same rates: $5 per million text input tokens, $8 per million image input tokens and $30 per million image output tokens. Resellers sell GPT Image 2 and 2.5 per image instead: on KIE it is $0.03 at 1K, $0.05 at 2K and $0.08 at 4K. GPT Image 2.6 stays token-billed and, in community tests, uses three to four times the output tokens of 2.5 at low and medium quality.

What resolution does GPT Image 2.5 support?

Each edge can go up to 3840 pixels, with total size between about 0.66 and 8.3 megapixels and an aspect ratio up to 3:1. OpenAI calls anything above about 3.7 megapixels experimental. Resellers offer 1K, 2K and 4K presets.

How do I keep my face the same in GPT Image?

Upload a clear, front-lit photo, number it as the subject, and write the change you want followed by what must not change: "Do not change the face, facial features, skin tone, hair, body shape or identity." Make one change per edit rather than five at once, and use medium quality or higher.

Does Starrd use GPT Image 2?

Yes. Since November 17, 2026 almost every Starrd template makes its first frame with GPT Image 2.5, then animates it into a video. Starrd's Custom photo mode also offers GPT Image 2.5 at 2K, 10 credits for one image or 30 for four.

Part of AI Video and Image Models, Compared

More guides in this series

Related Articles

Ready to create your own video?

Pick a template, upload your photos, and generate a cinematic AI video in minutes.

Browse Templates