Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities
Authors:
Gheorghe Comanici,
Eric Bieber,
Mike Schaekermann,
Ice Pasupat,
Noveen Sachdeva,
Inderjit Dhillon,
Marcel Blistein,
Ori Ram,
Dan Zhang,
Evan Rosen,
Luke Marris,
Sam Petulla,
Colin Gaffney,
Asaf Aharoni,
Nathan Lintz,
Tiago Cardal Pais,
Henrik Jacobsson,
Idan Szpektor,
Nan-Jiang Jiang,
Krishna Haridasan,
Ahmed Omran,
Nikunj Saunshi,
Dara Bahri,
Gaurav Mishra,
Eric Chu
, et al. (3410 additional authors not shown)
Abstract:
In this report, we introduce the Gemini 2.X model family: Gemini 2.5 Pro and Gemini 2.5 Flash, as well as our earlier Gemini 2.0 Flash and Flash-Lite models. Gemini 2.5 Pro is our most capable model yet, achieving SoTA performance on frontier coding and reasoning benchmarks. In addition to its incredible coding and reasoning skills, Gemini 2.5 Pro is a thinking model that excels at multimodal unde…
▽ More
In this report, we introduce the Gemini 2.X model family: Gemini 2.5 Pro and Gemini 2.5 Flash, as well as our earlier Gemini 2.0 Flash and Flash-Lite models. Gemini 2.5 Pro is our most capable model yet, achieving SoTA performance on frontier coding and reasoning benchmarks. In addition to its incredible coding and reasoning skills, Gemini 2.5 Pro is a thinking model that excels at multimodal understanding and it is now able to process up to 3 hours of video content. Its unique combination of long context, multimodal and reasoning capabilities can be combined to unlock new agentic workflows. Gemini 2.5 Flash provides excellent reasoning abilities at a fraction of the compute and latency requirements and Gemini 2.0 Flash and Flash-Lite provide high performance at low latency and cost. Taken together, the Gemini 2.X model generation spans the full Pareto frontier of model capability vs cost, allowing users to explore the boundaries of what is possible with complex agentic problem solving.
△ Less
Submitted 19 December, 2025; v1 submitted 7 July, 2025;
originally announced July 2025.
Murat: Multi-RAT False Base Station Detector
Authors:
Prajwol Kumar Nakarmi,
Mehmet Akif Ersoy,
Elif Ustundag Soykan,
Karl Norrman
Abstract:
In recent years, there has been an increasing interest in false base station detection systems. Most of these rely on software that users download into their mobile phones. The software either performs an analysis of radio environment measurements taken by the mobile phone or reports these measurements to a server on the Internet, which then analyzes the aggregated measurements collected from many…
▽ More
In recent years, there has been an increasing interest in false base station detection systems. Most of these rely on software that users download into their mobile phones. The software either performs an analysis of radio environment measurements taken by the mobile phone or reports these measurements to a server on the Internet, which then analyzes the aggregated measurements collected from many mobile phones. These systems suffer from two main drawbacks. First, they require modification to the mobile phones in the form of software and an active decision to participate from users. This severely limits the number of obtained measurements. Second, they do not make use of the information the mobile network has regarding network topology and configuration. This results in less reliable predictions than could be made. We present a network-based system for detecting false base stations that operate on any 3GPP radio access technology, without requiring modifications to mobile phones, and that allows taking full advantage of network topology and configuration information available to an operator. The analysis is performed by the mobile network based on measurement reports delivered by mobile phones as part of normal operations to maintain the wireless link. We implemented and validated the system in a lab experiment and a real operator trial. Our approach was adopted by the 3GPP standardization organization.
△ Less
Submitted 17 February, 2021;
originally announced February 2021.