Issues

• Feature request: deterministic verbatim utterance mode for fixed announcements

  NVIDIA/personaplex#107 · Filter by author JPMulder opened on Aug 22, 2026

• Service-Duplex-Bench

  question
  NVIDIA/personaplex#106 · Filter by author yangyyt opened on Aug 20, 2026

• --cpu-offload crashes at warmup: KV cache allocated on meta device for offloaded layers ("Cannot copy out of meta tensor")

  NVIDIA/personaplex#105 · Filter by author jordanice2026-bit opened on Aug 12, 2026

• No Retention of system prompt post the context window

  question
  NVIDIA/personaplex#99 · Filter by author tanmaylaud opened on Jun 19, 2026
  2

• The model cannot distinguish its role

  NVIDIA/personaplex#97 · Filter by author fitzwangnissan opened on May 22, 2026

• DGX Spark / GB10: PersonaPlex runs but local Web UI audio latency continuously increases

  bug
  NVIDIA/personaplex#96 · Filter by author frnzyyy opened on May 14, 2026
  1

• Feature request: typed user turns mid-session on /api/chat

  question
  NVIDIA/personaplex#95 · Filter by author apostololeg opened on May 9, 2026

• Smaller Checkpoint

  question
  NVIDIA/personaplex#94 · Filter by author Colin659 opened on May 8, 2026

• Tool call support?

  question
  NVIDIA/personaplex#93 · Filter by author tjasmin111 opened on May 7, 2026
  1

• Speaker option for this model

  question
  NVIDIA/personaplex#85 · Filter by author Colin659 opened on Apr 10, 2026
  1

• Bug found during model execution on Linux with CUDA (PyTorch backend)

  bug
  NVIDIA/personaplex#83 · Filter by author saiganesh47 opened on Apr 9, 2026

• remove israel name from the readme page

  NVIDIA/personaplex#74 · Filter by author sajin767 opened on Apr 7, 2026

• Spanish from Spain are too people

  question
  NVIDIA/personaplex#73 · Filter by author Gameadictive opened on Apr 7, 2026
  2

• MLX support for Apple Silicon

  NVIDIA/personaplex#71 · Filter by author BryanTegomoh opened on Apr 4, 2026
  1

• Voice clone

  question
  NVIDIA/personaplex#68 · Filter by author Sihan-Dong2007 opened on Mar 17, 2026
  2

• [Question] Details on negative-duration silence for barge-in/interruption simulation in training data

  question
  NVIDIA/personaplex#67 · Filter by author jungwon-choi opened on Mar 17, 2026
  1

• Connecting RAG and tool calls

  question
  NVIDIA/personaplex#66 · Filter by author Jank14 opened on Mar 12, 2026
  2

• RAG data

  question
  NVIDIA/personaplex#65 · Filter by author r1ckyb0nd opened on Mar 11, 2026

• Reason behind increasing dep_q from 8 to 16, compared to Moshi

  question
  NVIDIA/personaplex#57 · Filter by author siddharth1712 opened on Feb 5, 2026
  1

• I have created an 8-bit and 2-bit quantised version but it keep loading original

  question
  NVIDIA/personaplex#55 · Filter by author Skandu388 opened on Feb 5, 2026
  7

• Disabling Huggingface access causes failure

  bug
  NVIDIA/personaplex#54 · Filter by author TuxRobotics opened on Feb 4, 2026

• finetuning

  question
  NVIDIA/personaplex#52 · Filter by author Abhishek676062 opened on Feb 4, 2026
  1

• Floating Point 0 Error With H20

  bug
  NVIDIA/personaplex#51 · Filter by author khatyan opened on Feb 3, 2026
  1

• Excessive confusion and stupidity

  bug
  NVIDIA/personaplex#50 · Filter by author runvnc opened on Feb 3, 2026

• Clarification needed: Purpose of duplicate MimiModel instance (other_mimi) in server.py

  question
  NVIDIA/personaplex#46 · Filter by author asultanoff opened on Jan 29, 2026
  1