AI performance continues to increase, whilst prices continue to tumble. Today it's Google's turn...

Gemini Flash 2.0 just went GA. This is Google's smaller, cheaper model - except that it's actually better than it's (until today) top performing 1.5 Pro model. So Flash is the better model, but its pricing is under 1/10th of the price of 1.5 Pro. And Flash 2.0 is cheaper than the previous generation Flash 1.5.

To go along with Flash 2.0 we also get Gemini Pro 2.0, an upgrade to Google's top performing model. It's marked as "experimental", which means there's no pricing yet - in other words, it's free!

Not content with its low pricing of Flash, Google also announced Gemini Flash-Lite - which is cheaper than Flash. When we compare this model with OpenAI's cheapest, GPT4o-mini, Flash-Lite is actually half the price of GPT4o-mini. Not bad!

Let's not forget that Google, unlike everyone else, also provides a free tier with all its Gemini APIs - so for a lot of testing you're not going to pay a penny.

Almost every time we get a model announcement these days, we also get a cost reduction. Things are getting better and simultaneously cheaper.

Here's the announcement.