Perplexity’s latest round of updates is all about making the product feel more like a proper workhorse and less like a shiny demo. Over the past couple of weeks, the team has pushed improvements to model orchestration, automation, voice, coding, and research workflows, which means faster hand-offs, cleaner output, and fewer little fiddly steps for the rest of us. If you’ve been using Perplexity for campaign briefs, research, coding, or keeping a mountain of tabs in line, there’s a fair bit here to chew on.
Below, I’ve pulled together the most recent changes that show up in Perplexity’s changelog and related product updates from the last 14 days.
Quick note: the public release notes are a bit sparse in places, so I’ve only included updates that are clearly new or recently improved, based on the latest available announcements.
✅ Perplexity Computer is now available in Microsoft 365 apps
This update brings Perplexity Computer into Word, Excel, PowerPoint, Outlook, and Teams, which means you can work with it inside the tools a lot of people already live in all day. Instead of copying things back and forth between apps, you can keep the flow going where the work is actually happening.
For developers, that can mean checking specs while drafting docs in Word. For marketers, it can mean shaping a campaign brief in PowerPoint without hopping across ten browser tabs. For analysts, it’s handy for pulling together meeting notes in Teams and turning them into something usable before the coffee goes cold.
✅ Personal Computer is coming to Windows
Perplexity has been expanding its Personal Computer experience beyond Mac, with Windows support now on the way. That matters because it makes the product feel a lot less tied to one setup and a lot more ready for everyday office use.
If you’re a developer working on a Windows machine, this lowers the friction of using Perplexity alongside your usual stack. If you’re in operations or marketing, it means less messing around with platform workarounds and more time getting actual work done. Plain and simple, that’s one less hoop to jump through.
✅ Computer now behaves more like a full workspace, not just a chat box
The newest updates keep pushing Perplexity towards a more coordinated workspace, where it can handle planning, delegation, and delivery in a single flow. That makes it better suited to longer, messier jobs that usually bounce between tools, docs, spreadsheets, and browser searches.
For example, a content writer can draft a content plan, gather source material, and organise notes without starting from scratch three times. A marketer can use it to map out a launch checklist. A researcher can keep one thread going while the system helps manage the moving parts. It’s a bit like finally getting the farm gate to close properly after years of yanking it by hand.
✅ Custom Skills now automate repeating tasks
Perplexity has added Custom Skills, which are reusable workflows designed to handle repeatable jobs. In practical terms, this means you can package up a routine process once and use it again instead of rebuilding the same steps every single time.
That’s useful for developers setting up repeatable coding tasks, analysts generating regular summaries, and marketers producing the same type of campaign output week after week. If you’re always building similar briefs, summaries, or reports, this update saves time and cuts down on little human errors. Bless it, that sort of thing adds up fast.
✅ Model Council now compares multiple frontier models together
Perplexity’s Model Council update lets the system use multiple models in parallel and compare their responses. That gives users a more robust answer when a task needs nuance, checking, or a second opinion.
This is especially handy for research-heavy work, product analysis, and technical writing. A developer can compare different reasoning paths before shipping a solution. A strategist can sanity-check a positioning idea from more than one angle. A researcher can get a more balanced view when the answer is not black and white. It’s a smart way to keep one model from getting too full of itself.
✅ Voice Mode makes hands-free prompting easier
Voice Mode has been added so users can speak naturally instead of typing every request. That makes Perplexity feel a bit more conversational and a lot more useful when your hands are busy or you just want to move quicker.
Think of a marketer dictating campaign ideas while walking between meetings, or a founder capturing thoughts while juggling Slack, email, and a very needy browser. For analysts, it can be useful for quick follow-up questions while reviewing data. It’s the sort of thing that sounds small until you’re trying to work with one hand on a coffee and the other on a spreadsheet.
✅ Codex-powered coding support has been added for technical tasks
Perplexity has rolled out a dedicated coding sub-agent for tasks involving code, which means the system can hand off technical work more intelligently when a prompt calls for it. That makes coding help feel more targeted and less like a generic all-purpose response.
For developers, this can help with debugging, drafting code, and handling more structured technical prompts. For product teams, it can support quick prototype work. For technical marketers or analysts working with scripts, it can help tidy up code without making the user babysit every step. Handy as a pocket on a shirt, that one.
✅ GPT-5.4 and GPT-5.4 Thinking are now more widely available
Perplexity has expanded access to GPT-5.4 and GPT-5.4 Thinking, which improves reasoning, coding, and software interaction across the platform. The Thinking variant is designed for more step-by-step handling of tougher problems.
That matters for users who need stronger outputs on complex prompts, whether they are writing strategy docs, analysing data, or sorting through technical documentation. A developer might use it for trickier debugging. A writer could use it to sharpen structure and logic in a long draft. A researcher can lean on it when a question has a few too many moving parts for a quick skim answer.
✅ GPT-5.5 is now the default orchestration model in Perplexity Computer
Perplexity Computer has switched its default orchestration model to GPT-5.5, which is a behind-the-scenes change, but an important one. In plain English, it means the system is coordinating tasks with a newer brain, and that can improve how work is routed, planned, and completed.
Users may not notice the change in every single interaction, but they should feel it in smoother workflows and more reliable task handling. That’s especially relevant if you use Perplexity for long research chains, content production, or multi-step business tasks like syncing inventory data, summarising calls, or building internal reports.
✅ GPT Image 2 is now the default image model
Perplexity has also moved to GPT Image 2 as the default for image generation and editing inside Computer. That suggests a stronger focus on image quality and editing consistency for people who need visuals as part of their workflow.
For marketers, this can help with quick concept visuals or ad mock-ups. For content teams, it can support blog headers, social assets, and draft visuals without bouncing out to another tool. For product people, it can make it easier to spin up internal mock-ups or presentation-ready images without a big production detour.
✅ Direct Snowflake and Databricks data workflows are now supported
Perplexity has added direct data connections for Snowflake and Databricks, which is a solid win for teams living in warehouse-heavy environments. Instead of exporting data and dragging it somewhere else, users can work more directly with the source.
That’s especially useful for analysts preparing reports, operations teams checking metrics, and technical users who need cited outputs from live data. If your week includes pulling warehouse data for a board update or answering product questions from a dataset that never seems to sit still, this is a proper time saver.
✅ Reusable workflows and Space skills keep building out the automation side
Perplexity’s reusable workflows and Space skills point to the same direction: less rework, more repeatable output. These updates help users standardise the jobs they do often, rather than rebuilding the same process every time they need an answer or deliverable.
For a content writer, that might mean a repeatable research-and-outline workflow. For a marketer, it could be a campaign brief template that actually gets used. For a developer, it might be a structured way to investigate issues and document findings. That’s not flashy, but it saves a mountain of fiddly clicks.
Why this month’s updates matter
Across the board, Perplexity is leaning harder into being a practical AI work environment rather than just a search product. The newest changes make it easier to stay in one place, move faster across tasks, and trust the output a bit more when the work gets messy. If you’re juggling research, writing, data, and a dozen browser tabs that all look important at 4:15 on a Tuesday, these updates are worth a look.
Call to action: If you want to see what’s new for yourself, head over to Perplexity and give the latest features a spin. Share feedback if you’ve got it, and subscribe for future updates so you do not miss the next round of improvements. That’s how you keep the toolbox sharp, sugar.
