Issues

• Desktop: long Hugging Face model names are middle-truncated in the chat model picker

  raullenchai/Rapid-MLX#5664 · Filter by author raullenchai opened on Nov 7, 2026

• Desktop: model discovery hangs silently while macOS removable-volume permission is pending (HF cache on external drive)

  raullenchai/Rapid-MLX#5663 · Filter by author raullenchai opened on Nov 7, 2026

• [Tracking] Audit documentation usability on phone and narrow desktop viewports

  raullenchai/Rapid-MLX#5643 · Filter by author raullenchai opened on Nov 7, 2026
  3

• [Tracking] Recheck Homebrew vision dependency guidance from discussion #1205

  raullenchai/Rapid-MLX#5642 · Filter by author raullenchai opened on Nov 7, 2026
  2

• [Tracking] Triage remaining Desktop contributor requests from discussion #1763

  raullenchai/Rapid-MLX#5640 · Filter by author raullenchai opened on Nov 7, 2026
  2

• [Tracking] Revalidate GPT-OSS and Gemma agent-cache feedback from discussion #1205

  raullenchai/Rapid-MLX#5641 · Filter by author raullenchai opened on Nov 7, 2026
  2

• perf(qwen4): M5 tensor-unit path for QSA prefill

  P2performance
  raullenchai/Rapid-MLX#5618 · Filter by author raullenchai opened on Nov 5, 2026

• MTP: qwen4_exp draft head may not keep committed history (same class as #5571)

  bugengine/serverperformance
  raullenchai/Rapid-MLX#5591 · Filter by author raullenchai opened on Nov 4, 2026
  2

• MTP: hy3 draft head may not keep committed history (same class as #5571)

  bugengine/serverperformance
  raullenchai/Rapid-MLX#5590 · Filter by author raullenchai opened on Nov 4, 2026
  1

• why no translategemma model ?

  bugdesktop-appgood first issue
  raullenchai/Rapid-MLX#5589 · Filter by author ihubgit opened on Nov 4, 2026
  1

• serve: engine_start failures land in 'other' (70 coding-agent installs); qwen3.8-27b-4bit fails 25% of first starts on 32 GB and recovers on retry

  bugengine/serverP2telemetry
  raullenchai/Rapid-MLX#5578 · Filter by author raullenchai opened on Nov 4, 2026
  1

• Investigate report: Qwen3.8-27B 8-bit (BYOM) ~10 tok/s on M5 Max 128 GB

  benchmarkP2performance
  raullenchai/Rapid-MLX#5574 · Filter by author raullenchai opened on Nov 4, 2026

• Benchmark Qwen3.8-Flash-Next vs 27B at 64/96/128 GB and publish a per-tier recommendation

  benchmarkP2
  raullenchai/Rapid-MLX#5573 · Filter by author raullenchai opened on Nov 4, 2026

• Desktop/CLI: show expected tok/s for the user's chip before download and after first chat

  desktop-appenhancementP1ux
  raullenchai/Rapid-MLX#5572 · Filter by author raullenchai opened on Nov 4, 2026
  2

• Release runner loses model-cache access after reboot; cache gate reports it as missing snapshots

  bugci/releaseP1
  raullenchai/Rapid-MLX#5360 · Filter by author raullenchai opened on Oct 29, 2026
  2