MLX inference server for Apple Silicon — Text, Image, Video & Audio generation
JANG — Adaptive Mixed-Precision Quantization for Apple Silicon. v2.5.41 bounds DSV4 long-prefill pool attention: multi-token forwards past the dense-row cap get a tiled pool view (fixes Metal OOM past ~180K tokens) while decode keeps the dense BF16 fused-SDPA fast path.
Anthropic, PBC
Visionary sponsor
Bloomberg
Visionary sponsor
Hudson River Trading
Visionary sponsor
Meta
Visionary sponsor
NVIDIA
Visionary sponsor
Microsoft
Sustainability sponsor
Depot
Continuous Integration
AWS
Cloud computing and Security Sponsor
Datadog
Monitoring
Fastly
CDN
Google
Download Analytics
Sentry
Error logging
StatusPage
Status page