Google Gemini 1.5 Pro vs. Flash vs. Nano, Explained

To absolutely nobody’s surprise, AI took center stage at Google I/O, the company’s annual developers conference. And we got not one, but two new Gemini AI models, 1.5 Flash and Nano, along with some more information about the 1.5 Pro model announced in February.

The good news is both Gemini 1.5 Pro and 1.5 Flash are available in preview globally now. But if you want to try out Google’s new 2 million token context window, you’ll have to join the waitlist here.

Gemini 1.5 Pro and Flash are both natively multimodal, which means you can prompt them with text, images and videos. Google also announced new updates to the Gemini 1.5 API, including video frame extraction, parallel function calling and context caching — which, among other things, means you’ll only to upload your files to the models only once.

Gemini 1.5 Pro: The most advanced

Gemini 1.5 Pro was first released for early testing in February, where it had a maximum 128,000 token window. Now, it’s available more widely for developers and has a baseline token window of 1 million, with an option to sign up for a 2 million token window. 

• Cost: $7 per 1 million tokens; $3.50 per 1 million tokens for prompts up to 128,000
• Best for: Complex tasks, generating higher-quality results
• Response time: Slower than 1.5 Flash
• Token window: 1 million available now, 2 million available via waitlist

Gemini 1.5 Flash: Built for speed

Gemini 1.5 Flash is a new, leaner model introduced at Google I/O. Its main selling point is its response time, which is notably faster than 1.5 Pro. Gemini 1.5 Flash is available now globally. Here’s what you should know about it.

• Cost: Starting at $0.35 per million tokens
• Best for: Narrow and high-frequency tasks you need done ASAP
• Response time: Faster than 1.5 Pro
• Token window: 1 million available now, 2 million available via waitlist

Gemini Nano: Coming soon to a Pixel phone near you

Gemini Nano with Multimodality is going to be coming to Google Pixel smartphones later this year. The mobile AI model will be able to process text, images, audio and speech. Android was the first mobile operating system to include a built-in, on-device AI model, so this next update in the new Pixel will expand its AI capabilities.

During the I/O keynote, Dave Burke, vice president of engineering for Android, talked about how Nano on Pixels is going to be bringing some additional firepower to Android’s accessibility offerings. In particular, Nano will be part of updates to Android’s TalkBack feature, which provides auditory and text descriptions of images lacking alt text. He also showcased how a new Nano feature will help call out suspicious or spam phone calls.

Editor’s note: CNET is using an AI engine to help create a handful of stories. Reviews of AI products like this, just like CNET’s other hands-on reviews, are written by our human team of in-house experts. For more, see CNET’s AI policy and how we test AI.