What’s New in Perplexity: January 2026 Updates You Should Know

Look, I’ll be honest: keeping up with AI product updates feels a bit like trying to fold a fitted sheet. Just when you think you’ve got a handle on what’s new, three more features drop and you’re back to square one. But here’s the thing about Perplexity’s recent moves, they’re actually worth your attention. From new API offerings that promise faster responses at lower costs, to deeper integrations that make your research workflow feel less clunky, there’s some genuinely useful stuff happening. Whether you’re building something with their APIs, synthesising complex research, or just trying to get better answers faster, January’s shipped some solid improvements that make the platform feel less like a shiny demo and more like a tool you’d actually rely on.

New Sonar API Models: Speed and Smarts Without Breaking the Bank

In January, Perplexity rolled out two new API models called Sonar and Sonar Pro, and I reckon this is the update that’ll matter most if you’re a developer. The basic idea: Sonar gives you fast, straightforward answers. Sonar Pro tackles the gnarlier questions that need proper depth and pulls more sources for backing it up.

What actually matters here is that both come with built-in citations (no extra faffing about), automated rate limit scaling, and structured outputs available to everyone, not just paid tiers. The previous requirement that locked structured outputs behind a Tier 3 paywall? Gone.

Why this matters:

• Developers building chatbots or research tools get faster, cheaper infrastructure without sacrificing accuracy
• Content teams can automate research synthesis without juggling multiple API subscriptions
• Data analysts can pipe complex queries through structured outputs to feed straight into dashboards or reports

Pro Search: Multi-Step Reasoning Now Generally Available

Pro Search launched in beta ages ago, but as of November 2025 (and still going strong), it’s now generally available for Sonar Pro users. Here’s what it actually does: instead of just throwing a single search at your question, the model orchestrates multiple searches, fetches content from different URLs, and reasons through the whole thing to give you an answer that’s properly grounded.

The real-time thought streaming bit is genuinely useful too. You can watch the model’s reasoning process unfold as it works through your question. It’s like looking over someone’s shoulder while they research something, except they’re absurdly fast.

Who this helps:

• Researchers comparing conflicting sources without manually reading ten different tabs
• Marketers synthesising competitor analysis, industry trends, and audience insights in one go
• Analysts building reports that need layered, multi-source verification

Structured Outputs Democratised (March 2025 Update Still Relevant)

I keep coming back to this one because it’s quietly the most practical thing for people building on top of Perplexity. In March, they removed the paywall on structured outputs, JSON now works across all user tiers, all models. Before this, you’d need a higher tier to get your responses in structured format. Now you don’t.

If you’ve ever tried to pipe AI responses into your inventory system, your CRM, or a spreadsheet and had to manually parse responses, you’ll understand why this matters. Structured outputs mean the API returns clean JSON that your systems can actually consume without extra wrangling.

Practical applications:

• Auto-generate product comparison tables for e-commerce platforms
• Extract and format leads from market research into your CRM
• Parse complex industry reports into queryable data structures

Three New Search Modes: High, Balanced, and Fast

The latest Sonar model improvements added three new search modes. High for maximum depth on complex queries. Balanced for the middle ground. Fast when you just need something quick. The system can also auto-classify your query (using ‘search_type: auto’) and pick the right mode for you.

This is less flashy than other updates, but it’s the kind of thing that stops you from over-engineering your queries. You don’t need to spend two seconds getting a complex answer when a fast answer will do, and vice versa. It’s about matching effort to intention, which, I’ll admit, feels slightly contradictory when you’re used to just throwing everything at a search engine and hoping for the best.

Where this applies:

• Quick fact-checks during client calls (fast mode)
• Deep competitive analysis for strategy docs (high mode)
• General research workflows where you’re not sure how deep you need to go (auto mode)

iPad App: Actually Designed for Work Now

On January 16th, Perplexity redesigned and rewrote their iPad app from scratch to be properly optimized for iPadOS. I’ll be frank: lots of apps just scale their phone version up and call it a day. This one’s different. They built it for actual work, so split-screen research, multi-tab workflows, that sort of thing.

If you’re someone who does research or analysis on an iPad (and fair dinkum, more of us should be), this stops it feeling like you’re using a stretched phone app and actually gives you the real Perplexity experience on a bigger screen.

This helps:

• Researchers conducting multi-source synthesis on tablets
• Mobile workers who need proper research tools without a laptop
• Anyone who prefers iPad workflow for drafting and research alongside other tools

The Snapchat Integration: A Billion Users, Early 2026

This one’s less of a product update to Perplexity itself, more of a distribution play. Snap and Perplexity announced a partnership that lets nearly 1 billion Snapchat users ask questions and get verified answers inside the app. Perplexity is paying Snap 400 million dollars over a year for this.

Why I’m mentioning it: if you’re building content, marketing, or brand campaigns, this matters for reach. Perplexity answers will soon live in a platform where Gen Z and younger millennial users already spend heaps of time. It’s not just about Perplexity’s own platform anymore.

Practical implications:

• Brands can reach users seeking answers in a new, conversational context
• Publishers might see different traffic patterns as answers move into Snapchat
• Marketers need to think about how their content gets surfaced in citation-based answers across more platforms

What This All Means for You

Honestly, the throughline here isn’t one shiny new feature. It’s about Perplexity quietly making their infrastructure faster, cheaper, and more reliable for actual work. The API improvements mean developers can build less expensively. The iPad redesign means mobile workers get a proper tool. The search modes mean you’re not over-engineering every query. Pro Search still being available means complex research stops feeling like a slog through seventeen tabs.

I’m still figuring out how all of this fits into my own workflow, if I’m real about it. But that’s kind of the point. These aren’t updates that demand you rethink everything. They’re the kind that just make the thing work better when you actually use it.

Ready to Explore?

If you’ve been curious about Perplexity or you’re already using it, now’s a good time to dig into what’s shifted. Head over to perplexity.ai and poke around with the new features. If you’re a developer, the API docs are worth a proper read. If you’re a researcher or analyst, Pro Search might solve something that’s been nagging at your process. And if you’ve got feedback on what’s working or what’s not, Perplexity’s actually pretty responsive to users who give them real input. It’s a good moment to pay attention.