The mobile AI story has been elitist so far. Buy the expensive model, and you get on-device language models with faster AI tools. Buy something cheaper, and you mostly get a chatbot in the cloud.
However, MediaTek’s Dimensity 8550 could be the missing piece that changes that by pushing more advanced AI into the sub-flagship market.
The premium tier can create the hype, but Mobile AI only becomes mainstream when everyone can use it.
Quiz
Google Gemini
Trivia Challenge
From Bard to bold redesigns — how well do you know Google’s AI assistant and its evolving look?
OriginsDesignFeaturesBrandingAI Tech
What was the name of Google’s AI chatbot before it was rebranded as Gemini?
Correct! Google Bard launched in March 2023 and was rebranded to Gemini in February 2024. The name change reflected a shift to align the product with Google’s underlying Gemini large language model family.
Not quite — the answer is Google Bard. Bard launched in early 2023 as Google’s answer to ChatGPT, before being rebranded to Gemini in February 2024 to better reflect the AI model powering it.
The Gemini logo is characterized by a distinctive star-like shape. How many points does the Gemini logo mark have?
Correct! The Gemini logo features a four-pointed star that resembles a shimmering light or sparkle. This shape has become one of Google’s most recognisable AI iconography symbols across its product suite.
Not quite — the Gemini logo uses a four-pointed star shape. It’s designed to evoke a glimmering light or spark, and you’ll spot this same motif used across Google’s various AI-powered features throughout its apps.
What potentially controversial design change was Gemini recently spotted testing in its interface?
Correct! Gemini was spotted testing a design that relocates the chat input bar from its traditional bottom position to the top of the screen. This is considered controversial because it breaks from the ergonomic thumb-friendly convention most mobile chat apps follow.
Not quite — the controversial design change being tested is moving the chat input bar from the bottom to the top of the screen. While it may aid discoverability, many users find bottom-placed inputs more comfortable to reach on modern large-screen phones.
Which of the following best describes the Gemini model tier designed for the most complex, high-capability tasks?
Correct! Gemini Ultra is Google’s most powerful model tier, built for highly complex reasoning and multimodal tasks. It sits at the top of a tiered family that also includes Gemini Pro and the lightweight Gemini Nano.
Not quite — Gemini Ultra is the top-tier model. Google structured its Gemini family into tiers: Nano for on-device tasks, Pro for everyday use, and Ultra for the most demanding applications requiring deep reasoning and multimodal understanding.
Which Google Workspace app was one of the first to deeply integrate Gemini AI assistance directly into its interface?
Correct! Gmail was among the first Google Workspace apps to receive deep Gemini integration, offering features like email summarisation and AI-assisted reply drafting. This integration has since expanded across Docs, Sheets, and more.
Not quite — Gmail was one of the earliest and deepest Gemini integrations in Google Workspace. Features like smart summarisation and Help Me Write made it a flagship showcase for what Gemini could do inside everyday productivity tools.
In what month and year did Google officially rebrand Bard to Gemini?
Correct! Google officially rebranded Bard to Gemini in February 2024. The rebrand coincided with the launch of the dedicated Gemini mobile app and the introduction of the Gemini Advanced subscription tier powered by Ultra 1.0.
Not quite — the rebrand happened in February 2024. Google used the occasion to simultaneously launch the standalone Gemini app for Android and introduce Gemini Advanced, a premium subscription offering powered by the Ultra 1.0 model.
What is the primary color associated with the Gemini app icon and branding?
Correct! Gemini’s branding uses a flowing multicolour gradient blending blues, purples, and teals, evoking a cosmic, intelligent feel. This differentiates it visually from Google Assistant’s warmer colour palette and signals its AI-first identity.
Not quite — Gemini uses a multicolour gradient of blues, purples, and teals. The iridescent, almost galaxy-like quality of the palette was a deliberate design choice to set Gemini apart from previous Google AI products and give it a premium, futuristic feel.
What is the name of the lightweight Gemini model variant designed to run directly on-device, including on Android smartphones?
Correct! Gemini Nano is the on-device model designed to run locally on Android hardware without requiring a cloud connection. It powers features like Summarise in Recorder and Smart Reply, and was first deployed on Pixel 8 devices.
Not quite — it’s Gemini Nano. This compact model is optimised to run entirely on-device, meaning it can function without an internet connection. It debuted on the Pixel 8 series and enables privacy-friendly AI features that process data locally on your phone.
Your Score
/ 8
Thanks for playing!
Gemini Intelligence is too important to stay exclusive
Google laid out the bigger Gemini Intelligence vision before I/O. AI is becoming part of the phone’s plumbing, with local models and system-level access doing more work.
Rambler is a good example. It lives inside Gboard and cleans up speech as you talk. That is a big jump for voice-to-text technology, in my opinion.
The problem is that Google is keeping these features behind high-end hardware for now. Devices like the Google Pixel 10 and Samsung Galaxy S26 have confirmed native support for Gemini Nano V3, but other buyers are left out.
At some point, these tools need to be democratized. If features like this define the next era of Android, it makes no sense for them to stay locked to the most expensive phones.
My Pixel phone won’t get Gemini Intelligence, so I’m done with Google
Without Gemini Intelligence, my Pixel already feels outdated
The Dimensity 8550 is all about the NPU
MediaTek built the Dimensity 8550 for the next phase of mid-range phones. If you’re upgrading for faster gaming or a more responsive app UI, there is no difference between the two.
The main change is in the NPU, where MediaTek paired the existing NPU 880 with a dedicated Large Language Model Booster.
|
Dimensity 8500 |
Dimensity 8550 |
|
|---|---|---|
|
Fabrication |
TSMC 4nm (N4P) |
TSMC 4nm (N4P) |
|
CPU |
8-core Cortex-A725 |
8-core Cortex-A725 |
|
Peak CPU clock |
3.4GHz |
3.4GHz |
|
GPU |
Mali-G720 MC8 |
Mali-G720 MC8 |
|
NPU |
NPU 880 |
NPU 880 with LLM Booster |
The AI chip war is moving below flagships
MediaTek isn’t operating in a vacuum here. Phone makers are dragged into the AI arms race whether they like it or not.
Over in the Qualcomm camp, the Snapdragon 7 Gen 4 already shows the same direction of travel.
Qualcomm says its latest 7-series chip brings a 65% AI performance improvement over the previous generation, with an on-device Gen AI assistant and LLM support through its Hexagon NPU.
This competition seems like the industry at large has decided the sub-flagship tier is the next battlefield for local AI.
The chip alone won’t guarantee Gemini Intelligence
Having the right silicon isn’t enough on its own. Google set a 12GB RAM minimum for any phone trying to run Gemini Intelligence natively.
This is the catch for budget buyers. The Dimensity 8550 is not a magic pass. A phone maker can use the Dimensity 8550, pair it with 8GB of RAM to keep the price down, and still miss Gemini Nano V3 support.
Better AI could mean worse hardware elsewhere
There is no free lunch in smartphone manufacturing. If Google’s Gemini Intelligence mandates a strict 12GB of RAM, OEMs are forced to buy AI-heavy silicon for their next devices.
The bill of materials is a zero-sum game. If you are buying a phone in this tier, you are probably going to pay the “AI tax” one way or another.
To maintain aggressive price points, the next wave of flagship killers may regress in physical hardware quality. We might start seeing cheaper frames and “glastic” panels masquerading as premium materials.
The camera bumps are an easy target, too. While the primary sensor will likely remain a priority, it wouldn’t be surprising to see an increase in basic macro lenses or a downgrade in ultrawide cameras to balance the budget.
Google’s AI vision only works if everyone gets it
AI features only matter when they’re everywhere. Smart dictation and an autonomous task agent don’t reshape how phones are used if they only live on devices that cost more than a thousand dollars.
Now comes the harder part. MediaTek has opened the door, but we still need to see whether phone makers use it to build genuinely affordable AI phones, or turn the next wave of “flagship killers” into pricier devices with compromises elsewhere.
