MiniMax M2.7 is an open-weight sparse MoE LLM built for long-horizon, tool-rich agents. It integrates with NVIDIA platforms, NeMo fine-tuning and RL workflows, and open-source inference stacks for cost-effective deployment.
MiniMax M2.7 agentic AI model: Scalable Workflows on NVIDIA for Complex Applications Read Post »
A submissions policy at NeurIPS triggered a boycott threat from Chinese institutions, a rapid reversal, and a public apology. The episode shows how sanctions compliance and academic norms are colliding in global AI.
NeurIPS sanctions controversy: When AI research collides with geopolitics Read Post »
A 230-year-old industrial firm rolled out ChatGPT Enterprise to 650 staff with more than 125 custom GPTs, reporting 30–40% time savings on routine knowledge tasks and setting a path toward AI agents.
STADLER’s Case for ChatGPT Enterprise in manufacturing Read Post »
A federal judge paused the Pentagon’s supply‑chain‑risk label for Anthropic, halting related compliance moves while the case proceeds. Here’s what procurement and legal teams need to know right now.
MIT has launched a graduate program in Music Technology and Computation that centers on computational models of music, real-time interaction, and research spanning generative AI, acoustics, and audio signal processing. The work lays technical groundwork for translating sound into interactive forms and visuals.
Seeing Sounds: MIT’s New Program Points to AI-driven Sound Visualization Read Post »
MIT and Symbotic researchers report a hybrid controller that learns which robots to prioritize and then plans fast, safe trajectories around chokepoints. In simulations on real warehouse layouts, the approach lifted throughput by about 25% over strong baselines.
Warehouse robot traffic management AI moves from reactive to preventive control Read Post »
