What's the best Google AI?

Key Takeaways

Gemini 1.5 is the latest model from Google, using a new MoE architecture and larger token limit.
Gemini 1.5 Pro is best for general AI tasks with high accuracy, while Flash offers speed with less data.
Flash is faster but slightly less accurate than Pro, available to developers through AI Studio.

Artificial intelligence has reached its heyday, but with it comes a slew of new terms and variations. The latest version of Google’s AI platform Gemini (formerly called Bard) is Gemini 1.5. But, within Gemini 1.5, Google has developed two different tiers or sizes: Gemini 1.5 Pro and Gemini 1.5 Flash. But which is the better option? What’s the difference between Gemini 1.5 Pro and Gemini 1.5 Flash?

I tried NotebookLM with Gemini 1.5 Pro and it’s better than ChatGPT

I didn’t have to guess where the AI pulled information from, which made fact-checking and research so much easier.

The key to understanding Google’s different AI offerings is to break the name down into three parts. Gemini is the name for Google’s multi-modal AI. The 1.5 in the name refers to the version. The previous version was 1.0, but now Google has started rolling out 1.5. The final piece of the name is the tier or size of the AI model.

Google Gemini has four different tiers or sizes. Both the largest and the smallest tiers, Ultra and Nano, are still in version 1.0. Google released 1.5 to Pro first, followed by Flash. Google says Pro and Flash are the most-used sizes of Gemini, so the company rolled out the latest update to these two options first. The company hasn’t shared details yet about when (or if) Ultra 1.5 and Nano 1.5 will roll out.

Google refers to the Pro model as its “best model for general performance across a wide range of tasks.” Pro is the best tier available to general consumers, with the most accurate results. Flash, on the other hand, is a smaller model that packs in less data. That makes Flash deliver faster results than Pro, with slightly lower accuracy.

Currently, Gemini Advanced subscribers have access to Gemini 1.5 Pro. However, Gemini 1.5 Flash has currently only been released to developers inside AI Studio.

Here’s everything you need to know about the differences between Gemini 1.5 Pro vs. Gemini 1.5 Flash.

What is Gemini 1.5?

The latest model of Gemini is 1.5

Google Gemini image generation on an Android phone.

Gemini 1.5 is Google’s latest model of Gemini (as of June 2024), a multi-modal artificial intelligence platform. Google first introduced Gemini 1.5 with Gemini 1.5 Pro in February 2024.

Gemini 1.5 is different from Gemini 1.0 because it uses a new Mixture-of-Experts or MoE architecture. In the simplest terms, MoE means that the AI platform uses multiple training models, rather than a single one, and the predictions are combined in the final result. Each training model is considered an expert in a certain area, so mixing multiple “experts” is generally considered to improve the final result. The AI can also choose which expert to apply to a given prompt when all the training models are not required, which can lead to faster performance.

Gemini 1.5 is capable of handling longer prompts than 1.0.

The other major change with version 1.5 is a larger context window or token limit. This means that Gemini 1.5 is capable of handling longer prompts than 1.0. For example, Gemini 1.5 Pro can handle up to one million tokens, which is roughly an hour of video or over 700,000 words. That larger capacity comes into play when doling out tasks like asking AI to summarize an hour meeting, paraphrase a 70,000 word book, or analyze long stretches of code.

While the larger context window and MoE architecture are the biggest changes, 1.5 sees several other refinements, including improvements across modalities. Gemini is a multi-modal AI, which means it can accept more than just one type of input. In other words, Gemini can handle text inputs as well as photos, video and voice.

I added Gemini to my Chrome browser. Here’s how you can, too

Adding Google’s Gemini AI to my browser made for some efficient changes I didn’t even know I needed.

What is Gemini 1.5 Pro?

Pro is Gemini’s best blend of in-depth knowledge and speed

NotebookLM with Gemini 1.5 Pro in dark mode

Gemini 1.5 Pro is the first “size” of 1.5 available. The “Pro” part of the Gemini 1.5 Pro name indicates the size of the model.

AI platforms can very quickly become data and power hungry. To tailor to different needs, Google releases different tiers or sizes of the Gemini models. The larger sizes are meant for the most complex tasks, while the smallest model is capable of running on the device itself rather than relying on a connection to a data center housing all the AI infrastructure. For example, the highest tier, Ultra, is for developers running highly complex tasks, while the Nano tier is small enough to run the AI on a smartphone.

Pro is the second-highest tier of Gemini, one step down from Ultra (which is currently still in 1.0). Google calls this size the best general purpose AI across a wide range of tasks.

Gemini 1.5 Pro is now available to Gemini Advanced subscribers.

Google introduced Gemini 1.5 Pro exclusively to developers and enterprise users as a public preview. But, Gemini 1.5 Pro is now available to Gemini Advanced subscribers.

Gemini AI is Google’s new secret weapon against spam calls

Pixel phones are morphing into the bane of every phone scammers’ existence.

What is Gemini 1.5 Flash?

Gemini 1.5 Flash is the best choice for developing apps and chatbots that need big speed

Google

Flash is the tier that is one step below Pro. Google calls it the lightweight model of 1.5. Gemini 1.5 Flash is designed with speed and efficiency in mind. If speed matters, Flash is the model many developers will use.

Gemini 1.5 Flash is currently available as a public preview for testing, but only in Google’s AI Studio.

Project G-Assist: What we know about Nvidia's AI

What can Nvidia’s AI assistant, Project G-Assist, actually do for your games?

Project G-Assist’s biggest benefit is probably performance tuning. Here’s everything I could find out so far about Nvidia’s new gen AI.

What’s the difference between Gemini’s Flash and Pro models?

Only developers have access to Flash — so far

Google

First things first, Gemini 1.5 Flash has only been released to developers using AI Studio. Subscribers using Gemini Advanced don’t currently have the option to switch from the more advanced Gemini 1.5 Pro to the faster Flash tier yet.

The two tiers have the same one million token limit, which means both options can handle the same amount of data. The model that you choose won’t limit how much input you can put in, whether you are typing in one paragraph or uploading an hour long video.

Gemini 1.5 Flash will deliver results sooner than Gemini 1.5 Pro.

The key difference between the two is speed and accuracy. Gemini 1.5 Flash will deliver results sooner than Gemini 1.5 Pro. Google says that performance is similar between the two models. But, in benchmark tests, Gemini 1.5 Pro performed a bit better. The different scores between the two tiers on benchmark tests were minimal with tasks like image generation and language translation. Pro had a steeper point jump in areas like math and a lower error rate in automatic speech recognition.