OmniPkg is a distributed Python runtime hypervisor that enables concurrent, zero-copy multi-framework orchestration across multiple Python versions and toolchains.

Platform Support: - Linux: x8664, aarch64/ARM64, ppc64le - macOS: x8664 (Intel), arm64 (Apple Silicon) - Windows: x86_64, ARM64

Key Features: - Reproducible State Engine with deterministic environment lockfiles (export/sync exact states) - Native C-Dispatcher: sub-millisecond command routing that bypasses Python startup overhead - Microsecond package activation via warm daemon workers (~50μs) - Persistent model tagging: keep full models loaded in memory and route commands repeatedly without reloads or 5-minute timeouts - Support for Python 3.7 through 3.15 in a single environment (far beyond Conda’s typical 3.10–3.13/3.14 range) - Concurrent multi-version package and interpreter loading - True zero-copy tensor passing across different Python versions, PyTorch/CUDA versions, and frameworks — no JSON serialization or process boundaries required

Perfect for scientific computing, large model serving, and heterogeneous Python/CUDA workflows where maximum flexibility and minimal latency matter.

Note for noarch builds: - Core functionality works without the native C extension. - For maximum performance (sub-ms dispatching), install the platform-specific build instead.