OpenAI unveils GPT-5-Codex, a powerful AI coding assistant with dynamic thinking capabilities, outperforming previous models in agentic coding and code refactoring, while dynamically allocating 'thinking' time from seconds to hours on coding tasks.
OpenAI intensifies robotics research, hiring experts in humanoid robots and algorithms, with job listings signaling focus on teleoperation and simulation for robot training, accelerating pursuit of artificial general intelligence.
Google Gemini rockets to #1 on the US App Store, adding 23 million users in two weeks thanks to its free Nano Banana image editing model allowing up to 100 daily image generations for non-subscribers and 1,000 for paid subscribers.
Pioneering AI systems prioritize contextual awareness over raw computational prowess, leveraging intelligent information management for dynamic decision-making, enabling adaptive user engagement and elevated performance.
BentoML unveils llm-optimizer, an open-source tool enabling systematic benchmarking, automated parameter sweeps, and visualization of performance tradeoffs for optimizing self-hosted large language models, replacing ad-hoc trial and error with a constraint-driven, benchmark-focused approach.
Google AI unveils VaultGemma 1B, the largest private language model with differential privacy, achieving strong utility while preserving privacy, advancing safe and transparent AI models.
Anthropic releases Claude 4 Sonnet on Amazon Bedrock, superseding Claude 3.5, boasting increased context window, native reasoning mechanisms, and advanced tool use capabilities, requiring developers to update API calls, prompts, and application logic for compatibility.
Alibaba unveils Qwen-3-Max-Preview, a groundbreaking trillion-parameter language model, boosting AI capabilities with enhanced Chinese-English comprehension, instruction following, and complex query handling, available on Alibaba Cloud and OpenRouter at a premium price.
NVIDIA's TensorRT Model Optimizer introduces QAT and QAD techniques that recover AI accuracy lost during quantization by simulating low-precision arithmetic during training, with QAD leveraging a full-precision teacher model to guide quantized models, delivering benchmark improvements on models like Llama Nemotron Super via PyTorch and Hugging Face-compatible APIs.
Microsoft AI chief Mustafa Suleyman announces massive compute investments to build training clusters 6-10 times larger than current 15,000 H100 systems, while the company reportedly plans to integrate Anthropic's AI models into Microsoft 365 Copilot after they outperformed OpenAI models in Excel and PowerPoint.