Released MLX-LM v0.9.0 with enhanced server batching and Gemma 4 support

Snapdragon X2 Elite chip demonstrated power efficiency matching Apple M5 in benchmark test

Benchmark showing NPU 6-8x faster than CPU for OCR workloads

Leaked benchmark shows Apple's upcoming M5 Max NPU delivering 2x the inference speed of Intel's Panther Lake NPU on the Parakeet v3 model.

50th anniversary celebrations with executive appearances and potential AI service updates

Removed AI coding apps Replit and Vibecode from the App Store for violating guidelines

Apple's AI-powered Mac mini with Apple Intelligence features launched and sold out

Leak suggests upcoming Private Cloud Compute infrastructure for AI, built on clusters of four M2 Ultra chips.

Acquired Arduino (mentioned as occurring last year)

Launched Arduino Ventuno Q single-board computer for robotics and physical AI applications