Skip to content

EdgeFit AI v0.2.1 Preview

Pre-release
Pre-release

Choose a tag to compare

@Ron-Bee Ron-Bee released this 04 Nov 07:07
· 41 commits to main since this release

EdgeFit AI v0.2.1 Preview

EdgeFit AI preview release for the Meta Global AI Developer Hackathon. Includes hardware-aware model evaluation, on-device inference benchmarking, Meta-assisted recommendations, and evidence tracking.

Android physical testing was performed on debug builds. This specific signed release artifact has not yet completed independent physical installation validation.

An existing debug-signed installation may not accept an in-place update signed with a different certificate. Preserve existing app data before changing installations.

Artifact identity

  • File: EdgeFit-AI-Android-v0.2.1-preview.1-arm64.apk (8,567,890 bytes)
  • SHA-256: BC32EC0F61578F2CFE34258E4DB2CE671DFA8C8B452E84F70D26963D468C61E6 (also in SHA256SUMS.txt)
  • Package: ai.edgefit.diag, versionCode 3, versionName 0.2.1-preview.1
  • Signing certificate SHA-256: 0199C6DA4958AFD86E23BA27C90DD1E707A33F70934D70A38933DA2A920E6688 (RSA-4096, apksigner v2 verified)
  • minSdk 26, arm64-v8a, non-debuggable, INTERNET permission only
  • Source tag: v0.2.1-preview.1 (commit d4530f3)

Notes

  • GGUF model files are supplied separately by the user; the app never downloads models.
  • Meta features are BYOK (your own key, possible usage costs, explicit per-action consent).
  • 893/893 JVM tests green on release source; debug build verified 16/16 lifecycle tests on Galaxy S22 Ultra (SM-S908U, Android 16).
  • Not a Google Play release. No telemetry beyond user-initiated, consent-gated actions.

License and third-party notices

Original EdgeFit AI materials are rights-reserved (see LICENSE). Third-party components in this APK (llama.cpp/ggml, Kotlin runtime, LLVM OpenMP, Unicode data) remain under their respective licenses: THIRD_PARTY_NOTICES.md.