Explainer

GPT-6 Explained: Astra, Sol, Luna and What's New in ChatGPT

Short answer: GPT-6 is three models, and which one you get depends on your ChatGPT plan. Free users get Luna, Plus users get Sol with thinking, and only Pro, Business and Enterprise get Astra's top setting in chat. It's a big step for ChatGPT, but Claude Opus 5.5 still ranks higher on the independent index.

OpenAI rolled GPT-6 out in stages over five weeks. Astra arrived September 3, Sol and Luna on September 22, and on October 7 every ChatGPT user got GPT-6 plus a new way of answering called Intelligent UI. If you searched "GPT-6" and found three names, this page sorts them out.

A glowing glass chat bubble on a dark background with a bar chart, slider, toggle, calculator and checklist form unfolding out of it in teal light
Answers that build themselves: GPT-6's headline feature turns replies into charts, forms and small tools. Illustration: TechVerdict.
Covered GPT-6 Astra, Sol, Luna Claude Opus 5.5 Gemini 4 Argon

The three GPT-6 models

GPT-6 Astra
The flagship
$10 / $50 per 1M tokens
  • Powers the Pro thinking level in ChatGPT
  • 1.05M-token context, 128K output
  • In ChatGPT Work and Codex from Plus up

The catch: five times Sol's API price.

GPT-6 Sol
The everyday model
$2 / $10 GPT-6.1 Sol, per 1M
  • Runs paid ChatGPT, Instant to Extra High
  • Scores 52 on the independent index, one behind Astra
  • GPT-6.1 Sol is in Work, Codex and the API

The catch: 6.1 isn't in regular chat yet.

GPT-6 Luna
The free one
$0.10 / $0.50 per 1M tokens
  • Powers ChatGPT Free and Go
  • Gets Intelligent UI too
  • Cheapest OpenAI model for developers

The catch: scores 38, far below Sol.

GPT-6 in numbers

Infographic: GPT-6 Astra $10 input and $50 output per million tokens, index score 53; GPT-6.1 Sol $2 and $10, score 52; GPT-6 Luna $0.10 and $0.50, score 38; Claude Opus 5.5 leads at 58; key dates September 3, September 22 and October 7, 2026
Price and rank: API prices from OpenAI's pricing page, scores from the Artificial Analysis Intelligence Index, checked October 25, 2026. Infographic: TechVerdict.

What Intelligent UI does

Ask ChatGPT to split a dinner bill or plan retirement savings, and it can now answer with a working calculator instead of a paragraph. OpenAI says answers can include charts, tappable buttons, forms and small tools, built from ready-made parts that stream in while the reply is still being written.

The other change is speed. ChatGPT now starts answering while it is still thinking, and OpenAI says searched answers from GPT-6 Instant begin 44% sooner than on GPT-5.6 Instant.

OpenAI's GPT-6 with Intelligent UI art card: chat prompts answered with an interactive bike diagram and a football pitch diagram
The launch art: OpenAI's card for "GPT-6 and Intelligent UI for everyone," October 7. Image: OpenAI.
Too many charts? Intelligent UI has no separate quota, so it never eats into your limits. If you'd rather have plain text, turn down the Layout and visuals setting in ChatGPT. It also doesn't appear at the Pro thinking level or in the older Mac and Windows desktop apps.

Which ChatGPT plan gets which GPT-6

Plans and limits from OpenAI's pricing page and its "GPT-6 and other models in ChatGPT" help article, checked October 25, 2026.

PlanPriceChat modelThinking levelsAstra
Free$0Luna✗✗
Go$8/mo, may show adsLuna✗✗
Plus$20/moSolMedium and HighIn Work and Codex only
Pro$100, $200 or $500/moSolAll, plus Pro✓ GPT-6 Pro, weekly limits
Business$25/seat ($20 yearly)SolAll✓

OpenAI's pages disagree on the free model's version: its October 7 post says GPT-6 Luna, while its help article lists GPT-5.6 Luna.

The jump that matters is Free to Plus. Plus swaps Luna for Sol, adds thinking and drops ads. Pro is for people who hit Plus limits every week or want Astra's top setting. For the full plan breakdown, read Is ChatGPT Plus worth it?

How GPT-6 ranks

On the Artificial Analysis Intelligence Index, GPT-6 Astra scores 53 at its top setting, tied with Google's Gemini 4 Argon. Claude Opus 5.5 leads at 58. GPT-6.1 Sol sits one point behind Astra at 52 for a fifth of the price, which makes Sol the smart API pick.

OpenAI's own numbers are strong: 96.0% on GPQA Diamond, 72.6% on OSWorld 2.0 and 57.9% on Terminal-Bench 4.0. Its launch table quotes an older version of the index where Astra scores 61.2, so don't compare that with today's figures.

OpenAI's GPT-6 Astra launch art: a spiral of stars shaped like the number 6
The flagship: OpenAI's header art for GPT-6 Astra, September 3. Image: OpenAI.

The weak spots OpenAI admits

Should you use GPT-6?

You chat a few times a week
Free ChatGPT (Luna)
You use ChatGPT daily for work or study
Plus, $20/mo (Sol with thinking)
You build on the API
GPT-6.1 Sol at $2/$10
You want the top-ranked model
Claude Opus 5.5

The verdict

GPT-6 makes ChatGPT faster and more visual for everyone, and Sol is the model most people should care about. Plus is where GPT-6 starts to feel different, Astra is worth paying for only if you live in Pro-level work, and if you just want the strongest model, Claude Opus 5.5 still leads. For the head-to-head, see Claude Opus 5.5 vs GPT-6 Astra vs Gemini 4.

FAQ

When did GPT-6 come out?

OpenAI released GPT-6 Astra on September 3, 2026, then GPT-6 Sol and Luna on September 22. GPT-6 reached every ChatGPT user with Intelligent UI on October 7 for paid plans and October 8 for Free and Go.

Is GPT-6 free?

Partly. Free and Go users get the lightweight Luna model with Intelligent UI. The stronger Sol model and thinking levels need Plus ($20 a month) or higher, and the top model, Astra, runs the Pro thinking level on Pro, Business and Enterprise.

What is the difference between GPT-6 Astra, Sol and Luna?

Astra is OpenAI's most capable and most expensive model ($10/$50 per million tokens). Sol is the everyday model behind paid ChatGPT ($2/$10 for GPT-6.1 Sol). Luna is the small, cheap model behind free ChatGPT ($0.10/$0.50).

What is Intelligent UI in ChatGPT?

It lets ChatGPT answer with charts, buttons, forms and small tools, such as a bill splitter or a retirement calculator, that appear while the answer is still being written. It has no separate limit, and a Layout and visuals setting tones it down.

Is GPT-6 better than Claude Opus 5.5?

Not on the main independent index. Artificial Analysis scores GPT-6 Astra 53, tied with Gemini 4 Argon and five points behind Claude Opus 5.5 at 58. GPT-6 wins on reach: it is in ChatGPT for every user, including the free tier.