Download AnyInfer

v0.1.3

Every merge to main with a version bump produces a GitHub Release carrying the packages below, built by the release workflow.

Python Wheel

For applications. Published to PyPI; installing from the repository is byte-identical.

pip install anyinfer
# or, straight from the repository:
pip install "anyinfer @ git+https://github.com/anthturner/AnyInfer@main"

Latest release

Demo App Bundle

Self-contained: unzip and run, no Python required. Fully offline, no credentials needed.

Windows (x64)
macOS (Apple silicon) · macOS (Intel)
Linux (x64) · Linux (arm64)

Or with Python already installed:

pip install "anyinfer[demo]" && anyinfer-demo

Sidecar Bundle

The OpenAI-compatible service, self-contained for integrations that do not use Python.

Windows (x64)
macOS (Apple silicon) · macOS (Intel)
Linux (x64) · Linux (arm64)

Or with Python already installed:

pip install "anyinfer[serve]"
anyinfer serve --config anyinfer.json

Sidecar documentation

Notes

• The core wheel depends on only httpx2 and jsonschema; everything else is an extra. Installation and extras lists which ones you need.
• Unzip the demo bundle and run the anyinfer-demo executable inside; BUNDLE-INFO.txt records the exact version. Bundles never embed llama-server binaries or model weights; install a runtime explicitly and let the local subsystem acquire verified model artifacts when needed. (No 32-bit Windows bundle: PySide6 publishes no 32-bit Qt builds, though the pure-Python wheel runs there fine.)
• Unzip the sidecar bundle and run anyinfer-serve --config anyinfer.json from inside it. To keep it running across reboots, anyinfer-serve install writes the service definition for your platform after showing it to you (the archive's INSTALL.txt carries the same text); see running as a service. The bundle covers the HTTP frontend and the dependency-free adapters; integrations that need an optional provider SDK, such as GitHub Copilot or Entra authentication, should install the Python distribution with the corresponding extra.

Previous Releases and Checksums

The releases page keeps every prior version with its changelog, generated from the pull requests that landed between releases, plus the checksums for every attached asset. The versioning and branch policy is documented in the release strategy.