https://pypi.org/project/zmlx/ Recent updates to the Python Package Index for zmlx en https://pypi.org/project/zmlx/0.10.0/ ZMLX: Metal-kernel toolkit and optimization lab for MLX on Apple Silicon. Fused MoE decode (+2-12% on LFM2/Qwen3.5), custom GPU kernels in one line, 70+ kernel catalog. Tue, 03 Mar 2026 21:20:57 GMT https://pypi.org/project/zmlx/0.9.2/ ZMLX: Metal-kernel toolkit and optimization lab for MLX on Apple Silicon. Fused MoE decode (+6-12% on LFM2 8B/24B), custom GPU kernels in one line, 70+ kernel catalog. Wed, 25 Feb 2026 17:48:48 GMT https://pypi.org/project/zmlx/0.9.0/ ZMLX: Metal-kernel toolkit and optimization lab for MLX on Apple Silicon. Fused MoE decode (+6-12% on LFM2 8B/24B), custom GPU kernels in one line, 70+ kernel catalog. Tue, 24 Feb 2026 16:49:48 GMT https://pypi.org/project/zmlx/0.8.4/ ZMLX: Metal-kernel toolkit and optimization lab for MLX on Apple Silicon. Fused MoE decode (+5-12% on LFM2-8B-A1B), custom GPU kernels in one line, 70+ kernel catalog. Tue, 10 Feb 2026 19:45:12 GMT https://pypi.org/project/zmlx/0.8.3/ ZMLX: Metal-kernel toolkit and optimization lab for MLX on Apple Silicon. Fused MoE decode (+5-12% on LFM2-8B-A1B), custom GPU kernels in one line, 70+ kernel catalog. Sun, 08 Feb 2026 23:22:15 GMT https://pypi.org/project/zmlx/0.8.2/ ZMLX: Metal-kernel toolkit and optimization lab for MLX on Apple Silicon. Fused MoE decode (+5-12% on LFM2-8B-A1B), custom GPU kernels in one line, 70+ kernel catalog. Sat, 07 Feb 2026 20:01:38 GMT https://pypi.org/project/zmlx/0.8.0/ ZMLX: Metal-kernel toolkit and optimization lab for MLX on Apple Silicon. Fused MoE decode (+5-12% on LFM2-8B-A1B), custom GPU kernels in one line, 70+ kernel catalog. Wed, 04 Feb 2026 16:45:15 GMT https://pypi.org/project/zmlx/0.7.12/ ZMLX: Metal-kernel toolkit and optimization lab for MLX on Apple Silicon. Fused MoE decode (+5-12% on LFM2-8B-A1B), custom GPU kernels in one line, 70+ kernel catalog. Sun, 01 Feb 2026 22:20:37 GMT https://pypi.org/project/zmlx/0.7.11/ ZMLX: Metal-kernel toolkit and optimization lab for MLX on Apple Silicon. Fused MoE decode (+5-12% on LFM2-8B-A1B), custom GPU kernels in one line, 70+ kernel catalog. Sun, 01 Feb 2026 00:36:09 GMT https://pypi.org/project/zmlx/0.7.1/ ZMLX: Metal-kernel toolkit and optimization lab for MLX on Apple Silicon. Fused MoE decode (+5-26%), custom GPU kernels in one line, 70+ kernel catalog. Sat, 31 Jan 2026 20:44:11 GMT https://pypi.org/project/zmlx/0.7.0/ ZMLX: Metal-kernel toolkit and optimization lab for MLX on Apple Silicon. Fused MoE decode (+5-26%), custom GPU kernels in one line, 70+ kernel catalog. Sat, 31 Jan 2026 18:38:08 GMT https://pypi.org/project/zmlx/0.6.1/ ZMLX: Triton for Apple Silicon. Custom Metal GPU kernels for MLX with Triton-like ergonomics. Killer feature: 1.3-1.6x MoE inference speedup via fused expert routing. Fri, 30 Jan 2026 18:07:07 GMT https://pypi.org/project/zmlx/0.6.0/ ZMLX: Triton for Apple Silicon. Custom Metal GPU kernels for MLX with Triton-like ergonomics. Killer feature: 1.3-1.6x MoE inference speedup via fused expert routing. Fri, 30 Jan 2026 17:55:03 GMT https://pypi.org/project/zmlx/0.4.2/ ZMLX: Triton for Apple Silicon. Write custom Metal GPU kernels for MLX with one-line ergonomics, automatic gradients, and built-in autotuning. Fri, 30 Jan 2026 15:21:07 GMT https://pypi.org/project/zmlx/0.2.1/ ZMLX: Numba-for-MLX. Ergonomic helpers for authoring and autotuning MLX custom Metal kernels from Python. Fri, 30 Jan 2026 01:07:12 GMT https://pypi.org/project/zmlx/0.2.0/ ZMLX: Numba-for-MLX. Ergonomic helpers for authoring and autotuning MLX custom Metal kernels from Python. Thu, 29 Jan 2026 19:52:11 GMT