ChatGPT's Chat Tab Gets GPT-6 Sol and Interactive Answers Wednesday, With Luna for Free Users Due October 8
A.I. / news
ChatGPT's Chat Tab Gets GPT-6 Sol and Interactive Answers Wednesday, With Luna for Free Users Due October 8
OpenAI's Intelligent UI replaces some text replies with buttons, forms and charts. The speed and safety figures are OpenAI's own internal evaluations.

OpenAI began rolling out GPT-6 in the Chat tab of ChatGPT on Wednesday, October 7, adding a feature it calls Intelligent UI that lets the model reply with buttons, forms, charts and small interactive tools instead of text alone. Free and Go users are due to get the same update on Thursday, October 8, according to Unite.AI's report on the announcement.
OpenAI's post, titled "GPT-6 and Intelligent UI for everyone", appears in its news feed with a date of October 7. The page itself returned an access error when this report tried to load it, so the details below come from Unite.AI and TechCrunch coverage of OpenAI's statements. Unite.AI says its article was produced by an AI research agent and reviewed by its editors.
Which ChatGPT plans get which GPT-6 model
Plus, Pro, Business and Enterprise users get GPT-6 Sol in the Chat tab from Wednesday. Free and Go users get GPT-6 Luna from October 8. Enterprise access depends on workplace administrator settings, Unite.AI reported.
| Plan | Model in Chat | Date |
|---|---|---|
| Plus, Pro, Business, Enterprise | GPT-6 Sol | October 7 |
| Free, Go | GPT-6 Luna | October 8 |
The change applies only to Chat. The models behind ChatGPT Work and Codex are not changing in this release, Unite.AI said. Sol and Luna were first introduced on September 22.

What Intelligent UI does in ChatGPT
OpenAI built a library of native, streamable components and a compiler that renders the interface as the model generates it, according to Unite.AI. A plain text answer remains available when it is the better format.
The examples OpenAI gave were an interactive breakdown of a 7-speed bicycle with five selectable systems, and a lamb-roast plan with a shopping calculator that adjusts to guest count. Users can also ask for tools such as a savings-growth calculator, a dinner bill splitter or a playable game inside the chat.
OpenAI said work remains on the model's design judgment and on widening what it can build.
OpenAI's own speed and safety figures
OpenAI said GPT-6 Instant starts answering web-search questions 44% sooner on average than GPT-5.6 Instant. It also said GPT-6 Extra High began answering in the same time as GPT-5.6 Medium and scored higher than GPT-5.6 Extra High, on an internal evaluation of everyday agentic tasks.
Both figures are OpenAI's internal numbers. Unite.AI reported no independent verification, and OpenAI did not publish the task set in the coverage reviewed here.
TechCrunch reported on September 22 that the GPT-6 series costs half as much as the 5.6 series, which OpenAI attributed to improvements in caching and inference. It also reported OpenAI's claim that Sol makes about half as many mistakes as its predecessor in an internal factuality evaluation. OpenAI said both models outperform Anthropic's Fable and Opus, with no specifics.
OpenAI said ChatGPT has more than 1.2 billion weekly users, a figure Unite.AI repeated.
What the coverage does not say
No usage limit for Luna on the Free plan appears in either report. Neither gives per-token API prices for Sol or Luna, and the Enterprise rollout date depends on each customer's admin. OpenAI pointed to a system card for detailed safety results, which neither article reproduces.
The Decisions API is a separate OpenAI product with its own pricing. OpenAI's other recent claims, on maths, are covered in the site's Erdős Problems story. The Free and Go rollout is scheduled for Thursday, October 8.
Sources
More in A.I.
- 01LTX-2.5's Free Commercial Licence Stops at $10 Million in Group Revenue, and Its Card Asks for Contact DetailsLightricks' open-weight video and audio model is free for production use below that line, but the revenue test counts parent companies and affiliates, and the card publishes no benchmark scores.
- 02Gemini 4 Argon Costs $1.99 a Task at Promo Price Against $0.72 for GPT-6.1 Sol, and Is Not Yet on SaleGoogle's introductory $2 and $10 rates match OpenAI's Sol per token, but Artificial Analysis figures cited by eesel show Argon writing 62,000 output tokens a task where GPT-6 Astra writes 27,000.
- 03EmbeddingGemma 2 Embeds Text, Images, Audio and Video in 567MB of RAM on a Pixel 11 ProGoogle DeepMind's Apache 2.0 embedding model has 740M parameters in three modular pieces, and the only benchmark number its launch post prints is a 9.92-point gain on MTEB Code.
- 04Kolibri-1 Is Apache 2.0 and Fits on One B200, but Trails Qwen3.8 27B by 9.1 Points in GermanAleph Alpha's 78B-parameter mixture-of-experts model activates 3.46B parameters per token, and its own model card shows a larger dense Qwen ahead on every headline benchmark.