Gemini 1.5 Pro / Flash New Version Released!

Google's main AI lineup, Gemini, has officially released its previously experimental versions, 1.5 Pro Experimental and 1.5 Flash Experimental.Related article

With this release, previous limitations that prevented proper use due to low quotas have been lifted, and usage has been expanded to all regions, not just the US region. (At the time of writing, only us-central1 is available. Soon... right?)

In any case, the previous limit of only 10 uses per minute in us-central1 has been removed, expanding to all regions, and Flash's limit has been increased to a whopping 2000 RPM.

Price Reduction

As mentioned in the previous article, there have been significant performance improvements compared to the previous 1.5 Pro and 1.5 Flash, and this new release also includes a price reduction for 1.5 Pro.

Based on 1 million tokens, for inputs below 128K tokens:

Text Input is $1.25 -> $0.3125

Image Input is $0.001315 -> $0.00032875

The price has been reduced.

Safety Filter

It seems a new safety filter has been added, perhaps due to the US election period.

"Election-related" safety filter. When enabled, it filters out phrases that could influence the election...

Gemini 1.5 Flash 8b Exp

While not yet officially released, the existing "Gemini 1.5 Flash 8b Exp" model has been updated from 0827 to 0924. Personally, I had high expectations, and Google says the response has been quite positive.

(Naturally, because it's cheaper...)

It's currently only available in AI Studio, not yet in Vertex AI.

While performance is relatively lower due to the lower price, it's almost on par with the existing 1.5 Flash, with fast input/output speeds. Many people seem to have high hopes for it given its relatively low price.

For more details:

https://developers.googleblog.com/en/updated-production-ready-gemini-models-reduced-15-pro-pricing-increased-rate-limits-and-more/