Grok 4.5 is an xAI flagship model released on July 8, 2026. It's a reasoning model (it thinks step by step before answering) with a 500K-token context window and strong coding and agentic performance. It powered the Smart Grok family on Secret Chat until August 13, 2026, when the family moved to Grok 4.6.

Is Grok 4.5 still available on Secret Chat?

No. The Smart Grok family moved to Grok 4.6 on August 13, 2026, and only one Grok text model is offered at a time. You select the "Grok" model exactly as before — Smart Agentic for automatic routing, or a specific mode such as Web Search or Deep Research — and the answers now come from Grok 4.6.

What happened to Grok Fast?

It was retired on July 29, 2026. A quick tier needs a genuinely non-reasoning model, and that role now belongs to the free DeepSeek USA model, which is faster and stronger at reasoning and code. The Smart Grok family itself is unchanged in structure; it now runs Grok 4.6.

Can I turn off Grok 4.5's reasoning?

No — Grok 4.5 always reasons before answering, and neither does its successor Grok 4.6; reasoning can't be disabled. You can influence how hard it thinks (low/medium/high effort), and Secret Chat runs it at high effort by default. If you want a direct, non-reasoning answer instead, use the free DeepSeek USA model.

Is Grok private on Secret Chat?

Secret Chat forwards your prompts through its own proxy, so you use Grok (4.6 today, 4.5 while it was the current model) without an xAI account and without your requests being tied to you. No profile is built, no chat is associated with your identity, and your history stays in your browser. Provider-side retention may still apply, but your query reaches the model anonymized. Note that the content of your messages is sent verbatim — only your identity is removed.

Can Grok search the web on Secret Chat?

Yes, and this didn't change with the model. The Web Search and Deep Research modes attach xAI's live web and X search, so Grok combines its reasoning with current information instead of being limited to training-cutoff knowledge. That was true of 4.5 and is true of Grok 4.6, which runs the family today.

Can I have Grok's answer checked against the other models?

Yes; that's the AI Council. One question goes to models from two, three or four different companies at the same moment, each answer appears as that model finishes, and a referee then builds a disagreement table showing which model asserted each claim, which contradicted it and which never mentioned it, followed by a short synthesis of what at least two models reached independently. Grok is one of the four families available, so you can seat it in a two- or three-model panel of your choosing or run the Quartet with ChatGPT, Claude and Gemini. Because the models are queried in parallel, the check lasts about as long as its slowest member rather than the sum of the panel — far quicker than putting the same question to each of them by hand. Read it the right way round, though: agreement is evidence rather than proof, since models share training data and can be wrong together, and a council widens where your question travels, as it reaches every model on the panel plus the referee.

Grok 4.5 Is Now on Secret Chat AI: xAI's Reasoning Model

Published July 15, 2026 · Updated August 19, 2026

Update (August 13, 2026): Grok 4.5 has been superseded. The Smart Grok family on Secret Chat now runs Grok 4.6, announced by xAI on August 12, 2026. Everything below describes the Grok 4.5 release and is kept for reference; the structure it describes (Smart Agentic plus the manual modes) is unchanged, only the model behind it has moved on.

Good news for everyone who reaches for Grok: xAI's Grok 4.5 is now live on Secret Chat AI. The whole Smart Grok family (the automatic Smart Agentic mode and every manual mode behind it) has been upgraded to xAI's newest and most capable model, right inside the same private workspace where you already chat with GPT, Claude, Gemini and Perplexity.

Grok 4.5 is a genuine step up: a reasoning model that thinks before it answers, an Opus-class contender on the hardest coding and agentic tasks, and the current flagship of xAI's lineup. Below is what it is, exactly how it shows up on Secret Chat, the one place we deliberately kept the older Grok 4.3, and (as always) how you get all of it without a profile ever being built around you.

What Is Grok 4.5?

Grok 4.5 is xAI's latest frontier model, released on July 8, 2026. Unlike the fast, direct Grok models that came before it, Grok 4.5 is a reasoning model: it works through a problem step by step before it replies. A few things set it apart:

Reasoning is always on. Grok 4.5 always reasons before answering — reasoning can't be switched off. You can dial how hard it thinks (low, medium or high effort), and on Secret Chat it runs at high effort by default, so you get its deepest deliberation on every reply.
A 500K-token context window. It can hold roughly half a million tokens of context at once — long documents, sprawling codebases, and lengthy conversations stay in view while it works.
Opus-class capability. xAI positions Grok 4.5 at the top of its range, tuned hard for coding and long-horizon agentic work, with strong tool use, structured output and web/X search built in.

Grok 4.5 Powers the Whole Smart Grok Family

On Secret Chat, "Grok" isn't a single toggle — it's the Smart Grok model with an intelligent Smart Agentic router plus a set of manual modes you can choose yourself. As of this release, every one of those ran on Grok 4.5 (today they run Grok 4.6):

Smart Agentic — the default. A router reads each message and decides how to handle it: a quick reply for casual chat, a web search when you need fresh facts, or deep research for an involved question — now all answered by Grok 4.5.
Web Search — Grok 4.5 with xAI's live web and X (Twitter) search attached, so answers combine its reasoning with current information rather than training-cutoff knowledge.
Deep Research — the same model turned loose on multi-step research questions, gathering and synthesizing sources before it answers.
Image generation — Grok's image mode (Grok Imagine 2.0) for turning a prompt into a picture.

In short: pick the Grok model, and whether you leave it on Smart Agentic or switch to a specific mode, the brain behind it's now xAI's newest reasoning model.

Update: The Free Quick Model Is Now DeepSeek

When this article was published, the free quick-answer slot was filled by a separate Grok Fast model on the older, non-reasoning Grok 4.3. That model was retired on July 29, 2026.

The reasoning behind it still holds and is worth keeping: Grok 4.5 always reasons before replying and can't be told not to, so putting every throwaway question through a full reasoning pass would be slower and pricier for no benefit. A quick tier genuinely needs a non-reasoning model.

That role now belongs to DeepSeek USA, which is free for everyone — faster than Grok Fast was, considerably stronger at reasoning and code, and served only from US providers, never from DeepSeek's own Chinese API. See why open weights make that possible. The Smart Grok family below is unchanged and still runs Grok 4.5.

How to Use Grok 4.5 on Secret Chat AI

There's nothing to install or switch on — it's already live:

Open the Secret Chat app and select the Grok model in the model picker.
Leave it on Smart Agentic to let the router handle everything, or pick a specific mode next to the message box — Web Search, Deep Research, or image generation — when you already know what you want.

From there, every answer comes from the current Smart Grok model — Grok 4.6 since August 13, 2026. If you just want a fast, no-frills reply for a simple question, switch to the free DeepSeek USA model instead.

I want your best reasoning on a hard problem. Take your time, think it through step by step, weigh the approaches you would reject and why, and then give me your strongest answer. Here's the problem:

Private by Design, Whatever Grok You Pick

The upgrade changes the model, not the privacy. Secret Chat is a private gateway between you and the AI providers: your prompts are forwarded through our proxy, so you don't need an xAI account and your requests are never tied to you on the provider's side. We don't build a profile of you, and no chat is ever associated with your identity. Your conversation history lives locally in your own browser, not on our servers, and you can pay with crypto if you'd rather not link a card to your AI use.

That matters especially with Grok. As we covered in Does Grok Save Your Chats?, xAI trains on Grok conversations by default and keeps your history. Reaching Grok 4.5 through Secret Chat lets you use the exact same model as a stranger: retention may still apply at the provider, but your query arrives anonymized, not linked to your email or identity. One honest caveat, as always: Secret Chat removes you from your queries, it doesn't remove the content of your messages, so redacting any identifying details before you send is still up to you. For the full picture of how anonymized Grok access works, see our dedicated Private Grok page.

If you like having several models on tap for the same question, that's the whole idea. See why using GPT, Claude, Gemini, Grok and Perplexity together beats relying on any single one. Since this article was first published, the AI Council has turned that habit into one step, and Grok is one of the four families that can take a seat. The question goes out to models from several different companies at the same moment, and each answer opens in its own tab as that model finishes, so you're reading the first one while the rest are still being written, and the whole check runs for about as long as its slowest member rather than the sum of the panel. The manual version is the slow one: a tab per model, the same question pasted again in each, and every wait starting only once the previous one ended. You pick the size — a Duet of two models, a Trio of three, or the Quartet, which seats ChatGPT, Claude, Gemini and Grok together — and on the two smaller sizes you also pick which families take the seats, so Grok next to Claude and nothing else is an option if that's the comparison you want.

Then a referee model does the reading. Several independent answers to one question are several screens of prose that mostly agree, and the part worth your attention is the sentence or two where they don't — easy to skim past, since each model words it differently and buries it in a different paragraph. The referee pulls the factual claims out of every answer and lines them up model by model: who asserted each one, who contradicted it, who never raised it. What comes back is a finished conclusion: a short synthesis built only from claims at least two models reached independently, with whatever stays contested listed separately as things to verify before acting. What you have to read at the end is therefore shorter than one chatbot's reply, not longer, and every full answer stays on screen above it when you want to see what the summary rests on. Panels seat each family's mid tier rather than its heaviest, and when a question turns on current or checkable facts the whole panel answers with live web search attached (for Grok, the same xAI web and X search described above) decided once for the panel so the members stay comparable. Two limits we would rather state than imply: agreement is evidence, not proof, since models share training data and can be wrong together, which is what makes disagreement the reliable signal; and a council widens where your question travels, because 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.

Conclusion

Grok 4.5 was xAI's newest and most capable model when this article was published, and it powered the entire Smart Grok family on Secret Chat — Smart Agentic, Web Search, Deep Research and image generation alike. That family has since moved to Grok 4.6. The arrangement is the same: deep reasoning is reserved for the "Grok" model where it earns its keep, while the free quick-answer slot is filled by a genuinely non-reasoning model, today DeepSeek USA.

Try it today at Secret Chat AI, with the full multi-model lineup, your history in your own browser, and every query reaching the model anonymously.

Frequently Asked Questions

What is Grok 4.5?
Grok 4.5 is an xAI flagship model released on July 8, 2026. It's a reasoning model (it thinks step by step before answering) with a 500K-token context window and strong coding and agentic performance. It powered the Smart Grok family on Secret Chat until August 13, 2026, when the family moved to Grok 4.6.
Is Grok 4.5 still available on Secret Chat?
No. The Smart Grok family moved to Grok 4.6 on August 13, 2026, and only one Grok text model is offered at a time. You select the "Grok" model exactly as before — Smart Agentic for automatic routing, or a specific mode such as Web Search or Deep Research — and the answers now come from Grok 4.6.
What happened to Grok Fast?
It was retired on July 29, 2026. A quick tier needs a genuinely non-reasoning model, and that role now belongs to the free DeepSeek USA model, which is faster and stronger at reasoning and code. The Smart Grok family itself is unchanged in structure; it now runs Grok 4.6.
Can I turn off Grok 4.5's reasoning?
No — Grok 4.5 always reasons before answering, and neither does its successor Grok 4.6; reasoning can't be disabled. You can influence how hard it thinks (low/medium/high effort), and Secret Chat runs it at high effort by default. If you want a direct, non-reasoning answer instead, use the free DeepSeek USA model.
Is Grok private on Secret Chat?
Secret Chat forwards your prompts through its own proxy, so you use Grok (4.6 today, 4.5 while it was the current model) without an xAI account and without your requests being tied to you. No profile is built, no chat is associated with your identity, and your history stays in your browser. Provider-side retention may still apply, but your query reaches the model anonymized. Note that the content of your messages is sent verbatim — only your identity is removed.
Can Grok search the web on Secret Chat?
Yes, and this didn't change with the model. The Web Search and Deep Research modes attach xAI's live web and X search, so Grok combines its reasoning with current information instead of being limited to training-cutoff knowledge. That was true of 4.5 and is true of Grok 4.6, which runs the family today.
Can I have Grok's answer checked against the other models?
Yes; that's the AI Council. One question goes to models from two, three or four different companies at the same moment, each answer appears as that model finishes, and a referee then builds a disagreement table showing which model asserted each claim, which contradicted it and which never mentioned it, followed by a short synthesis of what at least two models reached independently. Grok is one of the four families available, so you can seat it in a two- or three-model panel of your choosing or run the Quartet with ChatGPT, Claude and Gemini. Because the models are queried in parallel, the check lasts about as long as its slowest member rather than the sum of the panel — far quicker than putting the same question to each of them by hand. Read it the right way round, though: agreement is evidence rather than proof, since models share training data and can be wrong together, and a council widens where your question travels, as it reaches every model on the panel plus the referee.