The reliable 2.5 Pro and incredibly effective 2.5 Flash-Lite versions of Google's Gemini AI family have been released.
Today, Google revealed a significant addition to its Gemini AI model family. The powerful Gemini 2.5 Pro is leaving preview and is prepared for developers to build upon after months of fine-tuning and adjustment. In the meantime, Google is giving a preview of the Gemini 2.5 Pro Flash-Lite, its next high-efficiency model. Despite its best efforts, Google is unable to avoid confusing model names.
The release of Gemini 2.5 in 2025 has raised Google's hopes for AI. Google is now more competitive with OpenAI and its well-liked GPT models thanks to these models' considerable improvements over earlier iterations. As Google moves closer to general availability, we have been deluged with previews and test builds, indicating that a model is stable enough for ongoing development.
Gemini 2.5 Pro trailed behind the 2.5 Flash model, which exited preview during I/O. Flash's 04-17 build goes into broad distribution today. As expected, the newly redesigned 06-05 build is the winner as Gemini 2.5 Pro exits preview and enters general release. This version seems to have resolved some of the problems that surfaced in the Google I/O build of 2.5 Pro.
Developers who need greater cost control will find all Gemini 2.5 models appealing because they come with configurable thinking budgets. Additionally, Google is launching Gemini 2.5 Flash-Lite, an experimental version, for the most cost-conscious developers. This paradigm, which is still in preview, provides an affordable means of handling large AI workloads. It is less than one-sixth of the price of output tokens and one-third of the price of text, picture, and video inputs when compared to 2.5 Flash. Since this version of Gemini is less powerful than 2.5 Flash and only makes sense when paying with tokens, it is unlikely to be added to the app for everyday users.