Google’s latest reasoning model, Gemini 2.5 Pro, achieved an 18.8% score on Humanity’s Last Exam without the need of any aids. It does extraordinarily well in reasoning, coding, science, and mathematics. On GPQA Diamond, Gemini 2.5 Pro scored 84%. Gemini 2.5 Pro is now being released to Gemini Advanced users. You may also try the new model for free in Google’s AI Studio.
Google has published a ground-breaking AI model named Gemini 2.5 Pro, which got 18.8% on Humanity’s Last Exam (HLE) without the use of web search or other resources. HLE is a rigorous benchmark developed by subject matter experts and prominent academicians from around the world to assess in-depth knowledge on a variety of topics. Previously, OpenAI’s o3-mini-high scored 14% on the same test without utilizing any tools.
Gemini 2.5 Pro is a thinking model, which means it’s a reasoning model built on top of a bigger base LLM that employs reinforcement learning and chain-of-thought prompting. Prior to the Gemini 2.5 Pro model, Google introduced the smaller Gemini 2.0 Flash Thinking variant. Google states that the Gemini 2.5 Pro model can “analyze information, draw logical conclusions, incorporate context and nuance, and make informed decisions.”
Google Gemini 2.5 Pro was tested on LMArena under the codename “nebula”. Gemini 2.5 Pro now leads the LMArena leaderboard with a score of 1,443, beating Grok 3 and GPT-4.5. In terms of other criteria, Google claims that Gemini 2.5 Pro excels at coding, math, and science. Gemini 2.5 Pro received an 84% GPQA Diamond score and an 86.7% AIME 2025 rating. Even in the SWE-verified assessment that assesses the capacity to solve real-world software issues, Gemini 2.5 Pro achieved 63.8%, trailing only Claude 3.7 Sonnet Extended Thinking, which scored 70.3%.
Google claims that the new Gemini 2.5 Pro model is capable of advanced coding and reasoning. It is being rolled out to Gemini Advanced users. Those who want to test the Gemini 2.5 Pro model for free should go to Google AI Studio (visit) and choose the “Gemini 2.5 Pro Experimental 03-25” model from the drop-down menu.
Your Own AI Helper That Enhances Gaming Is Nvidia G-Assist
Nvidia G-Assist enhances visuals, measures frame rates, and adjusts graphics settings. G-Assist also controls lighting for supported devices from Logitech, Corsair, MSI, and Nanoleaf. The capability is available in the Nvidia app and requires an RTX 30, 40, or 50 series desktop GPU with around 10GB of storage.
The Nvidia app has become one of the most useful software tools for every PC gamer due to its numerous usability features. One important reason for this is the app’s AI-powered functions, particularly G-Assist, which acts as a personal friend and optimizes your PC games.
Last year, Nvidia unveiled Project G-Assist, an AI gaming helper for all RTX-powered PCs. As part of its latest update, the Nvidia app now offers the G-Assist feature as an optional install. The new AI technology allows you to optimize game and system parameters in a single click. Furthermore, you may monitor frame rates and adjust the illumination on linked devices.
Nvidia G-Assist allows you to fine-tune your games.
Last year, we witnessed how G-Assist works, as the Nvidia showcase showed the entire RTX AI platform. Simply ask G-Assist: “Can you help me improve my framerate?” AI will respond with some metrics and solutions.
You may also request improved image quality settings for various devices and customize the performance to meet your specific requirements. This allows you to easily fine-tune graphics, video, and overall gaming settings. The AI also supports Logitech, Corsair, MSI, and Nanoleaf devices, allowing you to adjust the lighting on your system to your specifications. Furthermore, the AI functions as a co-pilot, which means that if you are having trouble understanding a game or its features, G-Assist will assist you. G-Assist provides thorough PC analysis and optimization, detecting frame rate limiters, GPU performance bottlenecks, and areas for development.
It also detects unsatisfactory display refresh rates and recommends remedies like GPU overclocking or resolution changes to reach the necessary frame rates. This AI assistant uses a local small language model and requires around 10GB of storage (6.5 GB for the System Assistant and 3 GB for Voice Commands) to function. Currently, Nvidia G-Assist is only available for RTX 30, 40, or 50 series desktop GPUs with at least 12GB of VRAM. Furthermore, the feature is now only available in English. To enable the feature, you must download and install the most recent Nvidia app update. Once done, press Alt+G to enable G-Assist. The Nvidia app received an update that included the new DLSS Override, Display Scaling, and Display Color Settings capabilities.
Download and install the latest Nvidia app update to take use of the new G-Assist feature. If you don’t already have the Nvidia app, get it from the button below and share your thoughts in the comments.