Google has added new hybrid reasoning AI models to its Gemini 2.5 family. According to the company, the Gemini 2.5 Pro and Gemini 2.5 Flash models are now widely accessible. It also unveiled a sneak peek at the new 2.5 Flash-Lite model, which it says is its fastest and most affordable model to date.
"We designed Gemini 2.5 to be a family of hybrid reasoning models that provide amazing performance, while also being at the Pareto Frontier of cost and speed," Google stated in its announcement.
Gemini 2.5 Pro and Flash go live
Google attributes the release of the generally available Gemini 2.5 Flash and 2.5 Pro versions for production applications to the insightful developer feedback it has received in recent weeks. Google has expanded its lineup with a sneak peek at the Gemini 2.5 Flash-Lite, billed as its fastest and most affordable 2.5-inch model to date.
"Gemini 2.5 Pro + 2.5 Flash are now stable and generally available. Plus, get a preview of Gemini 2.5 Flash-Lite, our fastest + most cost-efficient 2.5 model yet," Google CEO Sundar Pichai said in a post on X.
"Exciting steps as we expand our 2.5 series of hybrid reasoning models that deliver amazing performance at the Pareto frontier of cost and speed," he added.
According to Google, this new version offers lower latency than its predecessors, 2.0 Flash-Lite and 2.0 Flash, across a variety of prompts, making it ideal for high-volume, latency-sensitive tasks like translation and classification.
2.5 Flash-Lite maintains the essential features that characterize the Gemini 2.5 family in spite of its increased efficiency. The company claims that these include the capacity to modify computational "thinking" according to budget, integrate with programs like Google Search and code execution, support multimodal input (processing different kinds of data), and provide a significant context length of one million tokens.
Conclusion
Google claims that the model also exhibits "all-around higher quality" compared to 2.0 Flash-Lite in benchmarks related to multimodal tasks, coding, math, science, and reasoning.
Along with the recently stable versions of 2.5 Flash and Pro, developers can also access the preview of Gemini 2.5 Flash-Lite via Google AI Studio and Vertex AI. Additionally, the Gemini app now offers direct access to both 2.5 Flash and Pro. Additionally, Google Search now incorporates customized versions of Flash and 2.5 Flash Lite.
