cannitellinicholas-spec

cannitellinicholas-spec/PixelUnlockGPU

OpenAI-compatible on-device LiteRT-LM server for Pixel (Tensor G5) — serves Gemma to TypingMind and any OpenAI client over loopback, LAN, or Tailscale

5 good first / help-wanted issues · Kotlin · last activity Oct 2, 2026

0 stars 0 forks 0 watchers Kotlin Apache License 2.0
View on GitHub
5 Open Issues Need Help Last updated: Oct 2, 2026

Open Issues Need Help

View All on GitHub
Emit estimated usage block in chat completions (OpenAI clients expect it) about 1 month ago
enhancement good first issue
cannitellinicholas-spec/PixelUnlockGPU
0

OpenAI-compatible on-device LiteRT-LM server for Pixel (Tensor G5) — serves Gemma to TypingMind and any OpenAI client over loopback, LAN, or Tailscale

Kotlin
docs/api.md drifted from /health (access block, decode metrics) about 1 month ago
documentation good first issue
cannitellinicholas-spec/PixelUnlockGPU
0

OpenAI-compatible on-device LiteRT-LM server for Pixel (Tensor G5) — serves Gemma to TypingMind and any OpenAI client over loopback, LAN, or Tailscale

Kotlin
Auto-prefix KV reuse bails out when more than one trailing message is new about 1 month ago
enhancement good first issue
cannitellinicholas-spec/PixelUnlockGPU
0

OpenAI-compatible on-device LiteRT-LM server for Pixel (Tensor G5) — serves Gemma to TypingMind and any OpenAI client over loopback, LAN, or Tailscale

Kotlin
Client-side stop-sequence handling in the stream reassembler about 1 month ago
enhancement good first issue
cannitellinicholas-spec/PixelUnlockGPU
0

OpenAI-compatible on-device LiteRT-LM server for Pixel (Tensor G5) — serves Gemma to TypingMind and any OpenAI client over loopback, LAN, or Tailscale

Kotlin
Track Tensor G5 NPU backend (stock firmware abort) about 1 month ago
enhancement good first issue
cannitellinicholas-spec/PixelUnlockGPU
0

OpenAI-compatible on-device LiteRT-LM server for Pixel (Tensor G5) — serves Gemma to TypingMind and any OpenAI client over loopback, LAN, or Tailscale

Kotlin