Background

Google launched the Gemini Advanced plan on February 8, 2024, providing access to Google’s most powerful AI model Ultra 1.0. The Gemini Android App was also launched simultaneously.

After seeing the news, I immediately downloaded the Gemini Android App from the Play Store, but found the following error messages when opening:

• Domestic Vivo phone error message:

• Non-domestic Pixel 7 phone error message:

The above two error messages are from the February 8th first version of the Gemini Android App.

Solution

Method reference from V2EX post—Gemini app not working, anyone else experiencing this?

Using my neighbor’s Xiaomi HyperOS Android 14 phone as an example, their phone initially also showed device not supported error when opening Gemini App after downloading from US Play Store.

On their phone, downloaded Google Assistant App and Google App sequentially, entered Google Assistant App for initialization settings, but didn’t switch language as mentioned in the post [entered Google App settings in top right corner to switch language, but found the language column was blank].

Then opening Gemini App again, it worked normally.

Personal Experience

Yesterday on Chinese New Year’s Day (February 10th), after updating the Gemini App on my Pixel 7 phone, I found that as long as I switched the Google Assistant language to English (United States) on my Pixel 7, the app could run normally on my phone.

My domestic Vivo phone can’t be flashed to upgrade the Android system, so the Android version is still 10, which no longer meets the minimum system requirement for Gemini’s second Android App version [Android 12]. Maybe as Gemini Android App iterates, it will support lower Android versions.

Brief Review

After initial experience with Gemini Advanced, I found it’s still inferior to GPT-4. It confirms the saying that in AI, being one step ahead means staying ahead. Unless latecomers have a flash of inspiration, it’s hard to overtake in the black-box field of large language models.

After sharing a typical Gemini Advanced vs GPT-4 case I saw online, you’ll get an initial understanding of why I say Gemini Advanced is inferior to GPT-4.

Q: Today I own 3 cars but last year I sold 2 cars. How many cars do I own today?

A:

Gemini Advanced:

GPT-4:

Today I have 3 cars, but last year I sold 2 cars, how many cars do I have today? For this logic question, Gemini Advanced really gets confused by the cars sold last year, while GPT-4’s logic makes you feel like you’re talking to a real person.

The highlight that attracts me most about Gemini: Double-check response, using Google search to verify output.

Besides this highlight, nothing else really attracts me. Continue using GPT-4 peacefully and just enjoy the two months of free Gemini Advanced subscription.

Gemini backed by Google Search seems to have some advantages over GPT-4, but in reality is still outclassed by GPT-4 in most aspects. False advertising can’t beat users’ real experience. When OpenAI releases GPT-5 this year, it will surely be another dimensional strike against Google.

Competition between Google and OpenAI is ultimately good. With competition, the early realization of AGI becomes more likely.