Brave Software released a new stable version of the company’s Brave browser today. Brave 1.64 continues the AI push in the browser, but it also privacy improvements as well as new features.
Existing installations of the browser should receive it automatically via the integrated updating functionality. Desktop users may speed up the installation of the new version by loading brave://settings/help in the address bar. An alternative to that is to click on Menu > Help > About Brave.
Brave displays the current version when the page opens. It runs a check for updates and will download and install new updates automatically at this point. Note that it is necessary to restart Brave once to complete the installation.
One big change of Brave 1.64 deserves special mention. Brave notes in the changelog that it is no longer installing VPN services automatically. This is only done when users purchase a VPN subscription or enable it.
Tip: check out six Brave features that you may like.
Brave 1.64
Like Google and Microsoft, Brave Software is also integrating AI into its browser. Called Leo, it is different from Microsoft’s Copilot and Google’s Gemini. A core difference is that Leo gives you access to different large language models. Not all of them are free, but the core service is.
Brave 1.64 introduces a Leo AI Tools right-click context menu entry. Right-click on any text element on the webpage and you may select a diverse range of AI tools. The two main options are to ask the AI to provide context and for help with your creative efforts.
For instance, you could select a few paragraphs of text and ask Leo to summarize or explain it. When you are writing, you could ask Leo to rewrite the entire text, improve it, or change its tone or length.
How well this works depends on the text itself but also on other variables, including the large language model. Selecting the last sentence and asking Brave Leo to extend it resulted in the following three paragraphs of text:
The effectiveness of this method relies heavily on two main factors: the text being analyzed and the specific large language model being utilized. The text itself plays a crucial role, as the more nuanced and well-crafted it is, the better the language model can understand and generate appropriate responses. This, in turn, improves the overall outcome of the process.
However, the text is not the only determinant of success. The large language model chosen for the task also significantly influences the results. Different models have varying strengths and weaknesses, making some more suitable than others for specific tasks. For instance, a language model trained on scientific literature may not perform as well when dealing with colloquial language.
In conclusion, the combination of a well-structured text and a carefully selected language model is essential for achieving optimal results. By focusing on these two aspects, users can expect a more accurate and contextually appropriate response from the language model.
Other changes
The full changelog is long and you may want to check it out on Brave’s website to go through all the changes. Here is a quick overview of changes that are noteworthy:
- There is a new “Bring all tabs to this window” option in the right-click tab context menu. It moves all tabs of other windows into the active one.
- There is a new “show scrollbar” option when vertical tabs are enabled. This adds the option to use it for scrolling.
- Chromium’s storage partitioning is now compatible with Brave’s ephemeral storage implementation.
- Implemented stream isolation for third-party subresources in Private Windows with Tor to bring it up to par with Tor Browser’s implementation.
Now You: anything that you find interesting? Anything that I missed? Let me know in the comments.