Gemini 2.5 Flash Lite
Gemini 2.5 Flash Lite is Google’s cheapest multimodal model, designed for cost-sensitive workloads where latency and price matter more than peak quality.
Released
June 17, 2025
Type
multimodal
License
proprietary
Context
1,000,000 tokens
Pricing
Input
$0.07 / 1M tokens
Output
$0.30 / 1M tokens
Capabilities
Links
Gemini 2.5 Flash Lite in the news
MIT Technology Review · Jun 11, 2026
Google DeepMind is worried about what happens when millions of agents start to interact
Google DeepMind is funding research into the potential dangers arising from the interaction of millions of AI agents, as stated by Rohin Shah, director of the company’s AGI safety and alignment research. The increasing deployment of agents capable of performing tasks autonomously and following instructions from other agents introduces novel risks. To address this, Google DeepMind, alongside Schmidt Sciences, the UK government's ARIA agency, the Cooperative AI foundation, and Google.org, has committed $10 million to support research into multi-agent systems. This initiative aims to foster external academic research, as Shah noted that academia's strength lies in its ability to conduct long-term, forward-looking studies that might not be prioritized by industry labs. The primary goal is to establish a dedicated field of research for multi-agent safety, as currently, such a discipline is nascent. The concern is that a critical mass of interacting AI agents could lead to unforeseen consequences, mirroring how human institutions achieve complex outcomes beyond individual capabilities. Shah estimates that widespread agent deployment, posing significant risks, is still several months away.
MIT Technology Review · Jun 11, 2026
Job titles of the future: Nature’s drug designer
Tim Cernak, a former pharmaceutical chemist, transitioned his expertise to designing drugs for non-human patients in 2018, driven by a concern for ecosystem health. Cernak, now an associate professor at the University of Michigan, aims to develop pharmaceuticals specifically tailored for animals, contrasting with current practices where human medications are often used, leading to unintended harm. He has worked on treatments for various species, including frogs with fungal infections, Gila monsters with parasites, and bald eagles with avian flu. Cernak leverages artificial intelligence, specifically Google DeepMind's AlphaFold, to visualize protein structures and accelerate the drug design process. This AI tool allows for rapid generation of potential drug candidates that can bind to specific protein structures, a significant improvement over traditional methods. His lab utilizes robotic automation to test up to 1,500 potential drug compounds per day, drastically speeding up the experimental phase. Cernak's work also includes developing treatments for loggerhead sea turtles suffering from contagious tumors. He envisions a future where animal-specific drug development is the norm, ensuring more effective and less harmful treatments for wildlife.
Ars Technica · Jun 10, 2026
Google's latest DiffusionGemma open AI model comes with a 4x speed boost
Google DeepMind released DiffusionGemma, a new open AI model, on March 13, 2024, which generates text in parallel rather than sequentially. This approach, similar to image generation models, allows DiffusionGemma to produce an entire block of text at once, leading to increased speed and efficiency on local hardware. Unlike autoregressive models that generate text token by token, DiffusionGemma uses a denoising process over a field of placeholder tokens to refine its output. The model is a Mixture of Experts (MoE) with 26 billion parameters, but only 3.8 billion are active during inference, making it suitable for GPUs with 18GB of RAM. In benchmarks, DiffusionGemma achieved approximately 700 tokens per second on an RTX 5090 and over 1,000 tokens per second on a single Nvidia H100 AI accelerator. This performance represents a four-fold speed increase compared to similarly sized autoregressive Gemma models.