Zig + Metal homebrew engine runs Inkling 2 Mini faster than MLX on Mac
viksit · x · 2026-11-23
@rachpradhan's team wrote their own inference engine in Zig + Metal — no MLX, no Python in the model path — and ran Thinking Machines' Inkling 2 Mini faster than MLX and exo on the same Mac, thanks to their hand-rolled megakernel. Thinking Machines' Inkling-small-2 is also teased as coming soon.
More from Infra
- Inkling 2 Mini: a 26B local model that runs on laptops, and someone rebuilt Apple Notes around it — simonguozirui · 2026-11-23
- Agent-built sandbox platform hits ~1.2k creates/sec, built in a week by 1.5 engineers and 300+ agents — _arohan_ · 2026-11-23
- Context layers don't violate the bitter lesson: LLM-era indexes are data movement economics — CShorten30 · 2026-11-23
- WASM Linux Runtime author: reimplementing the kernel ABI, no emulation needed — OwariDa · 2026-11-23
- Dev Migrates From Vercel to Cloudflare in 35 Minutes With Claude, Cutting Bill From $70 to $10 — aloncarmel · 2026-11-23
- Same weights, one V100: inference engines differ up to 5x in tokens/s — PuzzleheadedArm3732 · 2026-11-23