Issues

• fix issues and bugs found by ArtLite eval

  Atomic-Germ/OpenFlowLM-Next#212 · Filter by author Cyronius opened on Oct 18, 2026
  2

• Granite 4.2 3B: closing the speed gap with llama.cpp Vulkan (prefill and decode)

  enhancementkernels
  Atomic-Germ/OpenFlowLM-Next#208 · Filter by author Cyronius opened on Oct 16, 2026

• openflowlm.org: what are the plans?

  Atomic-Germ/OpenFlowLM-Next#204 · Filter by author Cyronius opened on Oct 14, 2026
  2

  

• Request: ~9B NPU2 build for FLM 1.0.7 with MTP head retained (speculative decoding)

  Atomic-Germ/OpenFlowLM-Next#202 · Filter by author webactionhellas-dev opened on Oct 12, 2026
  6

• Granite 4.2 3B Uno drafter on the NPU (on top of #184)

  Atomic-Germ/OpenFlowLM-Next#189 · Filter by author Cyronius opened on Oct 9, 2026

• RFC: per-model and user settings — shipped defaults in model_list.json, a user file we never overwrite

  enhancement
  Atomic-Germ/OpenFlowLM-Next#188 · Filter by author Cyronius opened on Oct 9, 2026
  5

  

• Support decision models and System One API

  Atomic-Germ/OpenFlowLM-Next#158 · Filter by author danimalquackers opened on Oct 4, 2026
  1

• Run diffusion language models on the NPU: LLaDA2.2-mini and WeDLM-8B-Instruct

  engineenhancement
  Atomic-Germ/OpenFlowLM-Next#154 · Filter by author Cyronius opened on Oct 3, 2026
  1

  

• Run Qwen3.8-27B (dense) on the open kernels

  enhancementkernels
  Atomic-Germ/OpenFlowLM-Next#148 · Filter by author Cyronius opened on Oct 1, 2026 · 0.2.0
  1

• Image generation API: OpenAI-compatible /v1/images/generations in oflm serve

  enhancement
  Atomic-Germ/OpenFlowLM-Next#128 · Filter by author Cyronius opened on Sep 27, 2026 · 0.2.0
  

• Image generation on the NPU: install FLUX.2 [klein] 4B and run it with oflm image

  engineenhancement
  Atomic-Germ/OpenFlowLM-Next#127 · Filter by author Cyronius opened on Sep 27, 2026 · 0.2.0
  2

  

• oflm serve asks the OpenMP runtime to stop spinning and never gets it, which costs a third of prefill

  Atomic-Germ/OpenFlowLM-Next#106 · Filter by author Cyronius opened on Sep 21, 2026 · 0.2.0

• where does a llama.cpp backend live?

  Atomic-Germ/OpenFlowLM-Next#98 · Filter by author Cyronius opened on Sep 18, 2026
  11

• Finish GPT-OSS 20B on the open engine

  Atomic-Germ/OpenFlowLM-Next#91 · Filter by author Cyronius opened on Sep 15, 2026 · 0.2.0
  1

• oflm validate passes on machines where oflm run can't open the NPU

  Atomic-Germ/OpenFlowLM-Next#81 · Filter by author Cyronius opened on Sep 14, 2026 · 0.1.0
  2

  

• The 35B on open kernels: where the remaining gap is

  enhancementkernels
  Atomic-Germ/OpenFlowLM-Next#74 · Filter by author Cyronius opened on Sep 12, 2026 · 0.2.0
  4

  

• Three multimodal families still need the closed engine

  Atomic-Germ/OpenFlowLM-Next#73 · Filter by author Cyronius opened on Sep 12, 2026 · 0.2.0
  1

• A malformed JSON request body answers 500, not 400

  bug
  Atomic-Germ/OpenFlowLM-Next#69 · Filter by author vegah opened on Sep 12, 2026

• Add OTel Metrics

  Atomic-Germ/OpenFlowLM-Next#63 · Filter by author vegah opened on Sep 11, 2026

• open_npue: the two co-resident hw_context objects, measured — stock FLM only ever opens one

  Atomic-Germ/OpenFlowLM-Next#36 · Filter by author Platano78 opened on Sep 9, 2026
  3