Key launches and policy moves this week include Meta's Mac-ready open-weight model, Hugging Face's unified Granite Switch, OpenAI's academic access to GPT-Rosalind, Anthropic's transparency grant to METR, and Amazon Bee's new personal AI devices.
Meta and Apple released Muse Spark Open, an open-weight multimodal model that can run locally on Apple silicon devices.
Running large models on edge hardware reduces latency and data privacy concerns.
Source: TechCrunch AI
Hugging Face announced Granite Switch 4.1, a unified model offering chat, decision-making, and speech capabilities through a single API endpoint.
Unified models simplify integration and lower engineering overhead for multi-modal applications.
Source: Hugging Face
OpenAI introduced the Rosalind Academic Plan, granting eligible labs sponsored access to GPT-Rosalind and related tools for life-science research.
Provides researchers with powerful language models to accelerate scientific discovery without prohibitive costs.
Source: OpenAI
Anthropic granted the Model Evaluation and Transparency Research (METR) team unrestricted access to its models and internal data for independent risk assessments.
Greater transparency helps the community evaluate safety and bias of frontier AI systems.
Source: TechCrunch AI
Amazon-owned Bee released a wearable and a credit-card-sized recorder that can summarize conversations and generate content using on-device AI.
Personal AI hardware expands everyday AI interaction and creates new data collection points for developers.
Source: TechCrunch AI
Study how to deploy open-weight multimodal models on edge devices such as Apple silicon or low-power hardware. Understanding model quantization, hardware acceleration, and API integration will let you build applications that run locally, preserving privacy and reducing latency.
Build my Machine Learning Engineer plan