DeepSeek's V4.1-Flash carries 552B total parameters but activates only 8B per prompt, with native image input and up to a 1M-token context window. On OpenDesign Arena it scored 81.2/100, 98% of GPT-6 Astra's 82.7, at $0.023 per design versus Astra's $1.61.
OpenAI launched GPT-6 Astra, described as its most intelligent and aligned model yet, with state-of-the-art capabilities across computer use, coding, cybersecurity, and science. ChatGPT Work now runs on Astra, which OpenAI calls its best model for professional work.
iOS 27, macOS 27 Golden Gate, watchOS 27, visionOS 27 and tvOS 27 are out, with the LLM-based Siri AI as the flagship feature on all but tvOS. Apple's AFM 3 lineup includes a 3B-parameter on-device model and a 20B sparse model activating 1-4B parameters per request.
Two new GPT-Image-2.5 models: Sunburst for sharper details, natural lighting and precise edits, and Flare for high-quality images over 50% faster. Both support transparent backgrounds; rolling out to ChatGPT, ChatGPT Work and Codex users on desktop, mobile and web.