Google is expanding its Gemma 4 family with Gemma 4 12B, a new open model designed to bring powerful multimodal AI directly to laptops and everyday devices. The model sits between Google’s smaller edge-friendly E4B model and its larger 26B Mixture of Experts model, offering stronger reasoning and agentic abilities while keeping memory requirements low. Google says Gemma 4 models have now passed 150 million downloads, showing strong interest from developers building everything

OpenAI has surprised the mathematics world after one of its internal AI models found a counterexample to a famous problem first proposed by Hungarian mathematician Paul Erdős in 1946. The problem, known as the planar unit distance problem, asks a simple-sounding question: if a certain number of dots are placed on a flat plane, how many pairs of dots can be exactly one unit apart? For decades, mathematicians believed that grid-like arrangements were likely close to the best po

Google has officially launched Gemini 3 Flash, a faster and more cost efficient version of its Gemini 3 large language model, and is making it the default AI model across the Gemini app and Google Search’s AI Mode worldwide. Built on the same foundation as Gemini 3 Pro, Gemini 3 Flash is designed to deliver strong reasoning and multimodal performance while prioritizing speed, lower latency, and affordability. The release comes six months after Gemini 2.5 Flash and reflects

OpenAI is reportedly working on a new large language model codenamed “Garlic,” a move that signals just how quickly the competitive balance in AI can shift. According to reporting based on The Information, OpenAI’s Chief Research Officer Mark Chen told colleagues the model has been testing strongly inside the company, especially on coding and reasoning tasks, where it was described as performing well against Google’s Gemini 3 and Anthropic’s Opus 4.5. (Bakolia; “OpenAI De

Anthropic has introduced Claude Opus 4.5, a significant upgrade to its most capable AI model and one built specifically to excel at coding, professional workflows, and long-running agent tasks. Released shortly after updates to Google’s Gemini 3 and OpenAI’s GPT-5.1, the new model takes a different approach by focusing less on flashy content generation and more on dependable workplace performance (CNET). Anthropic says Opus 4.5 is designed for getting real work done, buildi