PageGrok and Page Assist: How to Choose

Published July 30, 2026 · UpdatedAugust 28, 2026

PageGrok and Page Assist are both open-source Chrome extensions that bring a model into the browser, and their centre of gravity differs. PageGrok is built around reading and asking about the page you are on, while Page Assist gives you a general-purpose chat workspace. This page starts with who each one suits, then gives a side-by-side table.

Who each one suits

PageGrok starts from the page in front of you. Opening the side panel reads the text already rendered on that page, so you ask about the content without pasting or tidying anything first. Built-in AI answers right after install with no model service to start, and once the extension fits how you work you can connect Ollama, LM Studio, oMLX, or any OpenAI-compatible endpoint. It suits people who read a lot of webpages every day and want as few steps as possible between installing and the first answer.

Page Assist starts from a chat window. It is also an open-source project, supports Ollama, Chrome’s built-in AI, and multiple OpenAI-compatible endpoints, and adds RAG and web search on top of chat. Before your first message you pick a provider and a model and get that connection working. It suits people who already run models locally and want a full-featured general front end.

Side by side

The details below were checked on 2026-08-28.

PageGrok Page Assist
Centre of gravity Reading and Q&A about the current page General-purpose chat workspace
How page content arrives Reads the active tab when you open the panel Chat with the current webpage from the sidebar
Before your first question Install, open the side panel, ask Configure a provider and a model
Model sources Built-in AI, Ollama, LM Studio, oMLX, OpenAI-compatible endpoints Ollama, Chrome built-in AI, multiple providers
Ask about part of a page Yes, clip one or several regions Works on the whole page
Beyond chat Suggested questions, multi-region clipping, conversation export RAG, web search
Open source Yes Yes

What PageGrok does for webpage reading

Everything in PageGrok is organised around the page you are currently on:

It reads the text already rendered in your active tab, which covers signed-in pages and content produced by front-end rendering. With the scope kept narrow, each step from parsing to asking to following up can be worked out in more detail.

When Page Assist is the better fit

Page Assist covers more ground in several situations:

If those describe your work, Page Assist is the simpler choice. The two extensions can also live side by side, each with its own button and settings: PageGrok while reading pages, Page Assist when you want a chat window that is not tied to one.

Install PageGrok →

Open the side panel after install and ask, with no signup. For Page Assist, its repository is the authority on features and supported providers; the details above are what we checked on 2026-08-28.

Further reading: Summarize any webpage, Local AI in the browser, Connect Ollama to your browser.

FAQ

Does PageGrok work right after install?

Yes. Open the side panel and it reads the text already rendered in your current tab, so you can ask about the content straight away. Built-in AI is the default path and starts with a free trial, with no account to create and no model service to start first.

Is PageGrok a fork of Page Assist?

No. They are independent open-source projects with separate codebases and different design goals. PageGrok is built around reading and following up on the current webpage, while Page Assist provides a general-purpose chat workspace.

Can I install both?

Yes. Each extension has its own toolbar button and settings, so installing both causes no conflict. A common setup is PageGrok while reading pages and Page Assist when you want a chat window that is not tied to a page.

What page content can PageGrok read?

It reads the text already rendered in your active tab, which includes signed-in pages and content produced by front-end rendering. It reads the page you already opened, so what it works from is exactly what you see.

Try it on the page you are reading right now

Built-in AI works with no setup and no account. You can also connect a local model.