Samsung Bixby And Perplexity AI In The New Gemini Era Of Phone Assistants

Samsung turns Bixby into a true AI partner with help from Perplexity

The smartphone AI race has taken a surprising twist. For years Apple has been mocked for moving slowly on artificial intelligence, while rivals rushed out flashy assistants and chatbots. Yet the overall architecture Apple is now building around Siri and Apple Intelligence is quietly becoming the template the rest of the industry follows. The latest example is Samsung, which is reportedly preparing to wire its Bixby assistant into Perplexity, one of the most capable AI answer engines on the market.

The information comes from well known tipster Semi-retired-ing on X, who claims that Samsung does not plan to rebuild Bixby from the ground up. Instead, starting with the upcoming Galaxy S26 series expected early next year, Bixby will act more like a traffic controller. Routine voice commands will continue to be handled by Samsung’s in house software, while heavier, more open ended queries are passed to Perplexity’s large language models in the background.

That division of labour mirrors the layered approach Apple uses. On iPhones, Apple Intelligence leans on small on device models for simple tasks such as setting reminders, editing messages or pulling up specific photos. When a request is too complex or too resource hungry, it can be offloaded to a powerful cloud model. Apple currently leans on OpenAI’s ChatGPT for some of those heavier requests, but has already laid out plans to switch to a customised version of Google’s Gemini in the months ahead.

Perplexity gives Bixby the brains it always needed

Perplexity built its reputation as an AI powered research tool that pulls in fresh information from the web, summarises it and surfaces sources in a conversational interface. It is much better suited to open ended questions than the rule based Bixby that launched with early Galaxy flagships. Rather than pouring years of investment into retraining its own models, Samsung is essentially bolting on a modern brain to an assistant that had fallen behind.

There is already a commercial relationship in place. Galaxy owners in the United States can claim up to twelve months of free access to Perplexity’s premium subscription tiers, a perk that made it clear Samsung sees value in the service. Extending that partnership so Bixby can quietly call on Perplexity in the background feels like a natural next step and a relatively low risk way for Samsung to refresh an assistant many users had started to ignore.

According to the early details, basic housekeeping requests will remain in Bixby’s lane: setting alarms, changing system settings, toggling connectivity and similar one line actions. Perplexity will be reserved for the more demanding jobs, including image generation, multi step reasoning, and complex planning queries that require understanding context over several messages. In practice, many users may barely notice which system is responding at any given moment, beyond seeing richer answers, better summaries and more creative outputs than Bixby could offer on its own.

This also gives Samsung another lever in its delicate relationship with Google. On Android phones today, Google’s Gemini assistant is deeply integrated and, by default, tends to dominate anything related to search or web based questions. By embedding Perplexity behind Bixby, Samsung can offer an alternative route for those queries and differentiate its Galaxy phones at a time when hardware innovation is slowing and software features are doing more of the marketing heavy lifting.

Apple’s next move raises the stakes for everyone

While Samsung looks to Perplexity, Apple is preparing a major upgrade of its own assistant stack. Reports earlier this month indicated that the company plans to run a gigantic, but heavily tailored, Gemini model in the cloud for the next generation of Siri. With around 1.2 trillion parameters, this customised Gemini is set to dwarf the bespoke 1.5 billion parameter model Apple currently uses in the cloud for Siri.

Parameter counts are not everything, but the jump from billions to well over a trillion represents a qualitative shift in what the system can handle. More parameters generally allow a model to capture more nuance, follow complicated instructions and juggle multiple pieces of context at once. For users, that should translate into Siri finally being able to manage long running conversations, complex automations and specialised knowledge queries that previously left it stumped.

Before settling on Gemini, Apple reportedly trialled OpenAI’s ChatGPT as well as Anthropic’s Claude models for handling these cloud level requests. In the end, the company opted for a customised flavour of Google’s technology that could be run inside Apple’s own Private Cloud Compute framework. That system is designed so that data sent to the cloud is stripped of personal identifiers, processed on hardened servers and never logged in a form that could be used to build advertising profiles.

Even so, the partnership underlines just how interdependent big tech rivals have become. Apple is expected to pay Google around one billion dollars per year for access to its proprietary AI models, adding another line item to a relationship that already sees Google paying roughly twenty billion dollars annually to remain the default search engine in Safari and across key Apple services. The two companies fight fiercely in some areas while quietly relying on one another behind the scenes.

A new phase in the assistant wars

Seen together, Apple’s Gemini powered Siri and Samsung’s Perplexity infused Bixby mark the beginning of a new phase in the voice assistant story. Instead of monolithic assistants trying to do everything themselves, we are moving toward orchestrators that hand off tasks to specialised large language models depending on complexity, privacy requirements and latency.

For Samsung, leaning on Perplexity may be the most efficient way to rescue Bixby from irrelevance and give Galaxy S26 buyers a clear AI headline feature without spending years rebuilding from scratch. For Apple, betting big on a massive Gemini variant is a way to keep control of the overall user experience while still tapping into the latest frontier models.

For users, the end result should be assistants that finally live up to a decade of promises. Whether you are talking to Siri on an iPhone or Bixby on a future Galaxy flagship, the real work will often be done by invisible models like Gemini, ChatGPT, Claude or Perplexity humming away in distant data centres. The only thing most people will care about is whether the assistant actually understands them and gets things done.