Which Model Do Free ChatGPT, Gemini and Perplexity Use? (October 2026)

Key takeaway: As of October 23, 2026, free ChatGPT answers with GPT-6 Luna, the free Gemini app uses Auto model selection (mostly Flash-Lite) and free Perplexity uses Best, which picks a model per query. GPT-6 Astra and Gemini 4 Argon, released in September and October, are paid-only, and no date has been announced for free users. The same question can surface different brands when a different model answers it: when RanketAI measured the same set of questions on the same day and changed only the model, its own brand mention rate was 43% on one model and 5% on the other. If you check whether your brand appears in AI answers, record which version produced each answer.

---

Three-line summary

• ChatGPT: since October 8, regular Chat on the Free and Go plans runs on GPT-6 Luna, and paid plans run on GPT-6 Sol. GPT-6 Astra is offered in the Work and Codex areas of paid plans.
• Gemini: since October 9, users without a plan default to Auto, and most questions are answered by Flash-Lite. Gemini 4 Argon is for Google AI Pro and Ultra subscribers.
• Perplexity: the free plan cannot pick a model; Best chooses one for each question. GPT-6 Astra and Gemini 4 Argon are on the Max plan list.

Free vs. paid models compared (as of October 23, 2026)

All three services give free users a fast, lighter model or automatic selection and keep their newest flagship models for paid plans.

Service	Free default	Lower / mid tier	Top tiers	Newest flagship
ChatGPT	GPT-6 Luna (Free and Go, regular Chat)	GPT-6 Sol (Plus, regular Chat)	GPT-6 Sol (Pro, Business, Enterprise, regular Chat)	GPT-6 Astra — Work and Codex on paid plans
Gemini app	Auto (mostly Flash-Lite)	AI Plus — Flash-Lite and Flash	AI Pro and Ultra — Flash-Lite, Flash and Pro	Gemini 4 Argon — AI Pro and Ultra
Perplexity	Best, automatic (no manual choice of advanced models)	Pro — GPT-6.1 Sol, Gemini 3.8 Flash, Claude Sonnet 5.5 and others	Max — the Pro list plus top models	GPT-6 Astra and Gemini 4 Argon — Max

The Gemini AI Plus cell follows press reports; Google only says AI Plus subscribers will receive an email explaining when the change applies to them. The Perplexity list is the help center snapshot as of October 17, 2026, and the help center says the model selector in your account is the source of truth.

Model names change often. Read this table as a dated record and recheck each app's model label at the time you run a check.

Free ChatGPT runs on GPT-6 Luna

Since October 8, 2026, regular Chat on ChatGPT's Free and Go plans has been answered by GPT-6 Luna, while paid plans use GPT-6 Sol.

OpenAI introduced GPT-6 Sol and Luna on September 22 in the Work and Codex areas for paid customers, then expanded the rollout to regular Chat on October 7. The community announcement described the models by plan:

"Paid tiers use GPT-6 Sol, while Free and Go use GPT-6 Luna." — OpenAI Developer Community announcement (2026-10-07)

The same post says the release only changes Chat and that the models in Work and Codex stay the same. The paid rollout began on October 7 and Free and Go followed on October 8 (Notebookcheck, citing OpenAI's release notes).

One caveat: according to Notebookcheck, OpenAI's model help page still listed GPT-5.6 Luna as the free default at that point. When official documents disagree, the more recent release notes are the better reference, and the model menu in the app is the most direct check.

GPT-6 Astra is OpenAI's new flagship model, released in early September. BleepingComputer quoted OpenAI: "GPT-6 Astra is now available to all Pro, Enterprise, and Business Premium users in ChatGPT Work and Codex," and reported that it would reach Plus and Business users gradually over a few days (BleepingComputer, 2026-09-06). No timeline has been announced for free users.

Free Gemini defaults to Auto, and most answers come from Flash-Lite

Since October 9, 2026, Gemini app users without a Google AI plan default to Auto model selection, and Google says most prompts go to Flash-Lite.

Some coverage described the change as "free users can only use Flash-Lite." Google's help article says something different: the default is Auto, and depending on the question, another model can answer.

"Auto will be the default setting for users without a Google AI plan. Gemini will use auto model selection to send each prompt to the Gemini model that suits it best. Most prompts will go to Flash-Lite, our fastest model." — Gemini Apps Help, Changes to Gemini model access and limits

According to the same article, if smart model selection is turned off, chats for users without a plan use Flash-Lite for every response. Users can also choose a low, medium or high effort level for each model. The article does not give version numbers.

Gemini 4 Argon is Google's new flagship model, made available in the Gemini app and API on October 13. Google's post that day said "Google AI Pro and Ultra subscribers can now access Gemini 4 Argon" (Google Blog, 2026-10-13). It does not mention the free or AI Plus plans.

The same change notice is published in the Korean-language help center with the same content, and it contains no country restriction. It applies to personal accounts; work and school accounts are outside the scope of that article.

Free Perplexity uses Best to pick a model per question

On Perplexity's free plan, advanced Search models cannot be chosen manually; Best automatically picks a suitable model for each question.

Perplexity's help center describes Best as an automatic model option inside Search rather than a separate mode. It explains that "Best picks a suitable model for each query so you do not have to," and that on the free plan "you cannot choose an advanced Search model manually" (Perplexity Help Center).

In the list dated October 17, 2026, the Pro plan can choose GPT-6.1 Sol, Gemini 3.8 Flash, Claude Sonnet 5.5 and others, while GPT-6 Astra, Gemini 4 Argon and Claude Opus 5.5 are Max-only. Because Perplexity runs a web search before answering any question, search results shape its answers relatively more than the model does. That structure is covered in Perplexity Never Skips the Web.

When the model changes, the brands in the answer can change too

The same question can surface very different brands depending on which model answers it, so record the model whenever you check brand visibility.

When RanketAI measured the same set of category questions on the same day in September 2026, changing only the model moved its own brand mention rate substantially (What Does a GEO Score Measure?).

Service	Model	RanketAI brand mention rate
ChatGPT	Model A	43%
ChatGPT	Model B (lighter model)	5%
Gemini	Version A	74%
Gemini	Version B (newer version)	40%

In ChatGPT, the gap moved together with search depth: model A averaged 4.3 web searches per answer and model B 1.2. This is one brand measured on one day, so the size of the gap should not be applied to other brands as is. The direction is clear, though: when a new model arrives, a brand's visibility in answers can shift.

With the October changes, the model answering free users changed on October 8 for ChatGPT and on October 9 for Gemini. When you compare records from before and after those dates, separate changes caused by your content from changes caused by the model switch.

Four steps to record the model when you check brand visibility

If you check brand visibility in AI answers yourself, log the service, plan and model next to each answer so that rounds can be compared.

1. Choose the version to check based on your goal. For consumer brands, the free version most customers use is the reference. If many of your customers are businesses that pay for these tools, check the paid version as well.
2. Record the model label for each answer. In ChatGPT, note the model menu; in Gemini, the model selector (write "Auto" if it is on Auto); in Perplexity, whether Best was used. With Gemini's Auto, a different model can answer each question.
3. Repeat each question at least three times. When automatic selection and normal answer variation overlap, one result is not enough. Record "mentioned in N of 3."
4. Mark model change dates in your log. Records from before and after a model change, such as October 8 (ChatGPT) and October 9 (Gemini), are not compared as the same condition.

How to choose questions and check sources is covered in How to Check If ChatGPT Mentions Your Brand: 5 Steps. If you need repeated measurement across several AI services under the same conditions, RanketAI's AI brand visibility analysis measures ChatGPT, Gemini and Perplexity answers together so you can compare mentions and sources answer by answer.

FAQ

Related reading

• What Does a GEO Score Measure? Prompts, Models, and Judges Change the Number — the original record of the model comparison.
• How to Check If ChatGPT Mentions Your Brand: 5 Steps — from preparing questions to repeat checks.
• 62% of Gemini Answers Came Without a Web Search — Korean SaaS Benchmark — how search decided source citation in Gemini.
• Measuring Brand Visibility in Gemini — the Mention-First Engine — why results differ by service.