Google has announced the stable release of Gemini 1.5 Pro and Gemini 1.5 Flash. There are 5 Pro models along with numerous API changes and enhancements to the Google AI studio. These updates will help the developers to create and deploy the AI applications at a faster rate and with lower costs.
Gemini 1.5 Pro
Gemini Pro 1.5 was first introduced to the public by Google for early testing in February. Note that Gemini is available in stable, it is also important to note that Google now has paid plans for Gemini in addition to the free tier that can be accessed through Google AI Studio.
For Gemini Pro 1. 5, users receive 2 requests per minute, 32000 tokens per minute, and 50 requests per day for free; your information is used to develop the product. Paid customers receive significantly higher rate limits, and their data is not processed.
Features of Gemini 1.5 Pro
- JSON schema mode is the latest feature introduced in the Gemini API and Google AI Studio earlier this year to help you manage the model output.
- To further increase the functionality available to developers in AI Studio, the light or dark UI mode or system defaults can now be set in the settings bar.
- Gemini 1. 5 Pro itself has several features like video frame extraction, parallel function calling, and context caching that are to be released in June.
Gemini 1.5 Flash
Gemini 1. 5 Flash was announced earlier this month at the Google I/O. It is a multimodal model designed for high speed with low latency and works best for narrow use cases.
Beginning June 17, Gemini 1. 5 Flash will also support model tuning for developers to fine-tune models for better optimization when deployed in production environments. Tuning will be available via Google AI Studio as well as the Gemini API with tuning jobs for now being free and no extra cost per token for using a tuned model.
Features of Gemini 1.5 Flash
- This model is well suited to high frequency, high throughput scenarios and so offers good cost savings where response time is paramount.
- With reduced weight, Gemini 1. 5 Flash, in general, exhibits good performance, especially in multimodal reasoning.
- Gemini 1. 5 Flash was designed specifically as the fastest and cheapest model for large-scale, high-volume work to accommodate developers’ requests for lower latency and cost.
Google has made both models more or less available across the world in over 200 countries and territories. These models are available through Google AI Studio where it is relatively cheap with flexible pay-per-use charges and hence can be used by developers as well as big organizations. These models are going to completely redefine the way developers and enterprises approach complicated tasks and big data thanks to their important characteristics such as extended context windows and multimodal functionality.