DeepResearch powered by Gemini 2.5 Pro!

DeepResearch powered by Gemini 2.5 Pro!

Gemini enhances its Deep Research feature by integrating the new 2.5 Pro Experimental model, which according to industry benchmarks and Chatbot Arena is considered the best performing AI model in the world - outperforming even OpenAI's DeepResearch! This extension is now available to all Gemini Advanced subscribers.

The Deep Research function, which has been upgraded with Gemini 2.5 Pro, shows significant improvements in analytical thinking and information synthesis. In internal tests, the reports generated by this system were favored by more than 2:1 compared to leading competitors.

The feature is available across all platforms and enables detailed, easy-to-understand research reports on virtually any topic - with significant time savings. Of particular interest is the new Audio Overviews feature, which converts reports into podcast-like conversations that can be listened to on the go.

Further testing will have to show how it really stacks up against OpenAI's state of the art DeepResearch, especially in Humans last exam. However, it must be said that given the speed at which improvements are occurring, a clear lead over OpenAI can be expected.

Llama 4 Maverick and Scout on ARC-AGI's Semi Private Evaluation.

Neuromodulation system reduces Alzheimer’s disease progression by 44%

A new, non-invasive neuromodulation system called SinaptiStim has slowed the progression of Alzheimer's disease by an impressive 44% in a Phase 2 clinical trial by specifically stimulating the brain's default mode network (DMN), offering a potentially groundbreaking treatment approach that goes beyond traditional biochemical interventions.

• Scientific breakthrough: The SinaptiStim system combines transcranial magnetic stimulation (rTMS) with real-time electroencephalography (EEG) and advanced imaging to deliver a personalized neuromodulation therapy that targets the brain's Default Mode Network (DMN), which becomes increasingly dysfunctional in Alzheimer's disease.

• Significance for aging research: This approach revolutionizes the understanding of Alzheimer's by viewing the disease not only as a biochemical problem but also as a disruption of neuronal network communication. The therapy preserves cognitive functions and slows decline, with patients retaining almost all abilities to cope with daily activities, while the control group showed significant deterioration.

• Role of AI: A machine learning system analyzes the EEG recordings of neural activity and generates personalized treatment protocols based on the patient's unique spatiotemporal patterns, enabling a precise, dynamic and adaptive therapeutic strategy.

• Societal benefits: The treatment showed statistically significant improvements in cognitive, functional and behavioral domains without serious side effects. It could enable the millions of Alzheimer's patients worldwide to maintain their independence and quality of life for much longer, while reducing the enormous socio-economic burden of care dependency.

• Future prospects: Sinaptica Therapeutics is already planning a larger Phase 3 study with 300 patients and is expanding its research to early Alzheimer's stages. This non-drug treatment option could represent a paradigm shift in neurodegeneration and may be applicable to other age-related neurodegenerative diseases, ultimately leading to healthier cognitive aging for society.

References

https://longevity.technology/news/neuromodulation-system-reduces-alzheimers-disease-progression-by-44/

https://sinapticatx.com

ChatGPT Web App Update Introduces New Model Variants: "o4-mini," "o4-mini-high," and "o3"

A new version of the ChatGPT web app was deployed just after midnight San Francisco time, bringing some notable backend updates. This release includes references to “o4-mini,” “o4-mini-high,” and “o3,” which likely indicate updates or additions to the suite of AI models powering the platform. While the specifics haven’t been officially detailed, these changes may reflect ongoing improvements to performance, efficiency, or customization options. Users might start noticing differences in how responses are generated or how the system handles various tasks.