The article reports that Mistral Large 4, a one‑trillion‑parameter multimodal model, is positioned as a European alternative to leading AI systems. OpenAI has launched its Decisions API in public beta, claiming it makes decisions up to ten times faster than GPT‑6 Luna and supports text and image inputs with three output types: predicates, choices, and scores, at a cost of $0.10 per 1 million input tokens. Google’s Gemini 3 series includes Nano Banana 2.1, a model based on Gemini 3.6 Flash that can take text and image inputs and output images and text, with a token context window of up to one million. The model is available through the Gemini app, API, Google AI Studio, Google Search AI Mode, Google Ads, Google Flow, and Google Stitch. The piece also notes that OpenAI has released a broad range of mathematical results from an internal frontier model, published in a GitHub repository with formalizations in Lean and summaries of reasoning, compute estimates, and problem statistics. It discusses the risks of open‑weight models, citing Chinese companies and GLM‑5.3, and mentions Google Project Zero’s guide on rapid vulnerability patching. Additional industry updates include DeepSeek’s $12 billion funding round, Claude’s integration with Google Docs, Sheets, and Slides, and the launch of an AI agent by Hark that handles everyday tasks. The article concludes with a brief on AI security events and a call to action for AI trust scoring tools.
Mistral Large 4 🧠, OpenAI Decisions API ❓, Nano Banana 2.1 🍌
Mistral Large 4, a one‑trillion‑parameter multimodal model, is positioned as a European alternative to leading AI systems. OpenAI’s Decisions API, now in public beta, offers faster decision making and supports text and image inputs. Google’s Gemini 3 series includes Nano Banana 2.1, a model with a 1‑million‑token context window that outputs text and images. The article also highlights AI security events, open‑weight model risks, and industry developments such as DeepSeek funding and Claude’s Google Docs integration.
Why it matters
The release of high‑capacity multimodal models and faster decision APIs signals rapid progress in AI capabilities, while the focus on security and governance highlights growing industry concerns about responsible deployment.