All pull requests

Pull requests

• [Fix] Add psutil to package dependencies

  #3555·arjunnai opened Sep 29, 2026·

  ·
  1

• [Fix] Check WebGPU per-dimension workgroup size limits

  #3553·JCNTH opened Sep 24, 2026·

  ·

• [Fix] Strip reasoning in history for Qwen3.5 chat templates

  #3550·shoemoney opened Sep 18, 2026·

  ·

• [Fix] Ignore stale callbacks in sync requests

  #3549·Pdevadiga45 opened Sep 12, 2026·

  ·

• [Embedding][Serve]Phase 2a: Add single-task embedding serve mode

  #3548·xthomaswang opened Sep 9, 2026Contributor·

  ·

• fix(ministral3): join the split log/error strings instead of passing them as extra args

  #3546·Anai-Guo opened Sep 6, 2026·

  ·

• feat: resolve oci:// models via llmman serve

  #3543·ericcurtin opened Aug 30, 2026·

  ·
  1

• Allow TVM FFI release candidates

  #3542·Marchematics opened Aug 30, 2026·

  ·
  1

• [Fix] argparse type=bool makes --disaggregation and --stream impossible to turn off

  #3541·Anai-Guo opened Aug 28, 2026·

  ·

• [Fix] Return 400 BadRequestError for malformed image URLs

  #3539·SuperMarioYL opened Aug 23, 2026·

  ·

• [Fix] Isolate callback state across engine instances

  #3538·zupengwang opened Aug 23, 2026·

  ·

• [Doc] Improve source build setup and validation

  #3535·akira-in-tech opened Aug 17, 2026·

  ·

• [Model] Enable support for GPT_OSS model

  #3533·krishnaraj36 opened Aug 7, 2026Contributor·

  ·

• fix(serve): copy request.stop so get_generation_config can't mutate it (#3512)

  #3531·Anai-Guo opened Aug 2, 2026·

  ·
  1

• [Docs] update installation guide with dependency and library path ins…

  #3527·farukerdem34 opened Jul 23, 2026·

  ·
  1

• [Chat] Add random-token benchmarking and token-count stats

  #3521·dsing-qual opened Jul 22, 2026·

  ·
  1

• [Model] Add Sarvam MoE model support

  #3519·dsing-qual opened Jul 22, 2026·

  ·

• [Docs] Recommend Python 3.13 for new environment setup (MacOS)

  #3518·farukerdem34 opened Jul 21, 2026·

  ·
  1

• [LoRA] Add an end-to-end single-adapter runtime path for Qwen2

  #3515·MagellaX opened Jul 17, 2026Contributor·

  ·
  3

• fix(protocol): check_function_call_usage raises raw KeyError on malformed tool_choice dict

  #3514·chuenchen309 opened Jul 15, 2026·

  ·
  2

• [Bug] Raise an actionable error when libmlc_llm shared dependencies are missing

  #3510·pjdurden opened Jul 8, 2026·

  ·
  6

• [Llama] Support attention_bias and mlp_bias config flags

  #3494·przemek-arch opened May 10, 2026·

  ·
  3

• Preserve Unicode text in generated mlc-chat-config.json

  #3491·mshr-h opened Apr 29, 2026Contributor·

  ·
  4

• [Embedding][Serve] Phase 4: Add first-class decoder embedding path

  #3488·xthomaswang opened Apr 23, 2026Contributor·

  ·
  12

• [Model] Add Gemma 4 E2B text-only support

  #3485·MakotoUwu opened Apr 19, 2026·

  ·
  7