Google has announced the update of the Gemini language model, both the Flash and Pro versions transforming into Gemini-1.5-Pro-002 and Gemini-1.5-Flash-002 respectively. A significant change is that both models have improved their test scores in almost every test set. In fact, Gemini-1.5-Flash-002 now outperforms Gemini-1.5-Pro-001 in several testing sets.
This update also includes a reduction in the price of Gemini Pro, both in input and output, along with an increase in the usage threshold to 2,000 RPM.
Google also reported that the response rate of Gemini Flash has steadily improved, reaching up to 300 token/s, while the latency has gradually decreased to as low as 300ms.
The version 002 models are now available for use in AI Studio and Vertex AI.
TLDR: Google has updated the Gemini language model, improving test scores and reducing prices for the Pro version while increasing usage thresholds. Gemini Flash shows improved response rates and reduced latency.
Leave a Comment