Secret Chat AI Now Runs GPT-5.6
Meet Sol, Terra, and Luna

Published July 11, 2026 · Updated August 19, 2026

We have upgraded. Secret Chat AI now runs OpenAI's newest model family — GPT-5.6, released on July 9, 2026 — across every GPT mode in the app. When you pick GPT, you're now talking to Sol, Terra, or Luna, the three tiers of the 5.6 generation. And the most important part hasn't changed one bit: you get these frontier models the same anonymized way you always have on Secret Chat.

Here's what the new family is, how Secret Chat uses each tier, and why the upgrade changes what the models can do without changing what they know about you.

What Is GPT-5.6?

GPT-5.6 is OpenAI's latest generation, and it comes with a new, cleaner naming idea: the number is the generation (5.6), and the name is the capability tier. Instead of a soup of "nano / mini / pro" suffixes, you get three named tiers:

• Sol — the flagship, and the strongest, most capable model in the family. It leads OpenAI's benchmarks for the release and unlocks the new maximum reasoning effort and an "Ultra Mode" that lets the model spin up its own subagents to work through hard, multi-step problems.
• Terra — the balanced, everyday tier. OpenAI positions it as matching the previous generation (GPT-5.5) on quality while being about twice as cheap to run, the sensible default for most work.
• Luna, the fast, low-cost tier. It's the quickest and most affordable model in the family, built for snappy answers where you don't need the flagship's deliberation.

Update — the cheap tiers got much cheaper. On 30 July 2026, three weeks after launch, OpenAI cut Luna by 80% to $0.20 per million input tokens and $1.20 per million output, and Terra by 20% to $2.00 and $12.00. Sol didn't move, at $5.00 and $30.00. Since your credits convert to what the model actually costs us, that cut reaches you directly: a Fast Chat answer on Luna now consumes roughly a fifth of the credits it did in mid-July, and Web Search on Terra about four-fifths.

Alongside the models, OpenAI introduced a more agentic direction (an assistant it calls "ChatGPT Work"), and developer improvements like programmatic tool calling and better prompt caching. For everyday Secret Chat users, the headline is simpler: the three tiers are meaningfully more capable, and the cheaper tiers are faster and cost less to run than before.

How Secret Chat Uses Each Tier

You don't have to learn the tiers — Secret Chat maps them onto the GPT modes you already use, and Smart Agentic picks the right one for you automatically. Under the hood:

So a quick question gets Luna's speed, an everyday task gets Terra's balance, and a hard problem gets Sol's full reasoning, and in Smart Agentic mode the model works out which of those your message needs, then does it. You can still pick a mode by hand whenever you want to steer.

Terra has a second job outside the GPT modes. The AI Council sends one question to several models from different companies at once — a Duet of two models, a Trio of three, or a Quartet running ChatGPT, Claude, Gemini and Grok together — reveals each answer the moment that model finishes, and then has a cheap referee model extract the factual claims into a disagreement table: which model asserted each claim, which contradicted it, which never mentioned it, plus a short synthesis of what at least two models agreed on and a list of things to verify before acting. ChatGPT's seat on those panels is the balanced everyday tier rather than the flagship, and that's deliberate: a council is evidence because independent companies disagree, not because each individual answer is the deepest one obtainable. Depth from a single model is what Research and Sol are for; whether several companies agree is what a council is for.

Two limits we would rather state than imply. Agreement is evidence, not proof: models share training data and can be wrong together, so disagreement is the reliable signal. And a council widens where your question travels, since it reaches every model on the panel plus the referee. Each request is still anonymous, but anonymization protects who you are, not what you type. The finished check is stored only on your device, and the Session Privacy Report lists every model that took part.

You Get the Newest Models: Still Fully Anonymized

Upgrading the models doesn't change the thing that makes Secret Chat what it is: it's an anonymizer and depersonalizer. Its whole design is no profile, no association, no training — it makes you unlinkable to OpenAI, so you use GPT-5.6 as a stranger.

• No profile of you. Registration takes an email — used only for account access and payment, never to store or associate your prompts with you, with no name, phone number, or behavioral portrait built up across your chats.
• No chat associated with you. Secret Chat reaches GPT through OpenAI's business API on your behalf, so OpenAI sees the gateway, not you; no identity, no account, and no IP address attached to any prompt.
• Never used for training. Under those business-API terms your prompts aren't used to train OpenAI's models, and Secret Chat sends an explicit deletion request per session, recording the outcome in a Session Privacy Report (PDF).
• History only in your browser. Your conversations live in local storage on your device, not a cloud archive under your identity, and you can switch between GPT-5.6 and other private AI models in one place, or have the AI Council run GPT next to Claude, Gemini and Grok on one question, with a referee comparing the answers claim by claim.

This is why a model upgrade is a pure win here: even where OpenAI's API keeps data for its short retention window, what is kept is not linked to you. The one honest boundary is unchanged too — Secret Chat AI is an anonymizer, not a content filter. OpenAI's servers still have to read your prompt to answer it, and we don't encrypt or strip the words you send. It removes who you are from the request, not what the request says, so keep names and identifiers you wouldn't want a model to see out of your prompts.

Do You Need to Do Anything?

No. The migration is already live and automatic: the same interface, the same email-only account, the same local storage, and the same credit-based billing. Open Secret Chat, pick GPT, and you're on GPT-5.6. If you were mid-project, nothing about your saved chats or files changes; only the model answering you got better.

Frequently Asked Questions

1. What are Sol, Terra, and Luna?

   They're the three tiers of OpenAI's GPT-5.6 family, released July 9, 2026. Sol is the flagship (most capable, with maximum reasoning and Ultra Mode), Terra is the balanced everyday tier (around twice as cheap as the previous generation), and Luna is the fastest, lowest-cost tier.

2. Which one does Secret Chat use?

   All three. Fast Chat uses Luna, Web Search uses Terra, and Research uses Sol, and Smart Agentic mode picks the right tier for your request automatically. You can also choose a mode by hand.

3. Which GPT-5.6 tier answers in the AI Council?

   The balanced everyday tier, not the flagship. The AI Council puts ChatGPT next to models from other companies — a Duet of two, a Trio of three, or a Quartet with Claude, Gemini and Grok — and seats each family's mid tier by design, because the evidence a council produces comes from independent companies disagreeing rather than from any single answer being the deepest obtainable. Use Research and Sol when you want one model's deepest work, and a council when you want to see where several companies contradict each other on the same question.

4. Does the upgrade change my privacy?

   No; it stays exactly as strong. Secret Chat reaches GPT-5.6 through OpenAI's business API anonymously: no profile, no account or IP attached, your email (used only for account access and payment) never associated with your prompts, and prompts not used for training. OpenAI still processes each prompt to answer it, so remove identifiers you wouldn't want it to see.

5. Do I need to update or reconfigure anything?

   No. The migration is automatic and already live. Your account, chats, files, and credits are unchanged — only the underlying model improved.

Try It

The upgrade is the easy kind: better models, same privacy. Open Secret Chat AI, choose GPT, and put GPT-5.6 to work — anonymously, with your history on your own device and a choice of models a click away.

Sources

• OpenAI — Previewing GPT-5.6 Sol
• OpenAI Help Center — A preview of GPT-5.6 (Sol, Terra, Luna)