All pull requests

Pull requests

• Add replay fixtures for NVIDIA response-shape parser coverage

  #384·vidushig-nv opened Oct 9, 2026Collaborator·

  ·
  2

• Guard model-registry removal proposals by catalog completeness

  #380·Jon Patterson (freshyjmp) opened Oct 5, 2026Collaborator·

  ·

• fix(trt): update urllib3 to 2.8.0

  #379·John Kennedy (jkennedyvz) opened Oct 5, 2026Contributor·

  ·

• fix(langgraph): make LangGraphExtractor.extract use the source it is given

  #371·Ch Revanth reddy (revanth-045) opened Sep 28, 2026·

  ·

• feat(ai-endpoints): add nemotron 3.5 super vl capabilities

  #370·Sean Lopp (slopp) opened Sep 28, 2026Contributor·

  ·

• fix(openshell): update vulnerable ai-endpoints dependency

  #369·John Kennedy (jkennedyvz) opened Sep 28, 2026Contributor·

  ·

• fix(ai-endpoints): don't mutate the global MODEL_TABLE in available_models

  #350·Andrew Chen (chuenchen309) opened Jul 11, 2026Contributor·

  ·

• fix(ai-endpoints): don't mutate the caller's message in _nv_vlm_adjust_input

  #348·Andrew Chen (chuenchen309) opened Jul 11, 2026Contributor·

  ·

• fix(ai-endpoints): honor per-call stream_options instead of double-passing it

  #347·Andrew Chen (chuenchen309) opened Jul 11, 2026Contributor·

  ·

• fix(ai-endpoints): accept a CA directory for verify_ssl on the async client

  #346·Andrew Chen (chuenchen309) opened Jul 11, 2026Contributor·

  ·

• fix(ai-endpoints): keep text before <think> when parsing reasoning

  #345·Andrew Chen (chuenchen309) opened Jul 11, 2026Contributor·

  ·

• fix(ai-endpoints): correct misspelled trucate embeddings arg to truncate

  #344·Andrew Chen (chuenchen309) opened Jul 11, 2026Contributor·

  ·

• fix(ai-endpoints): add stacklevel=2 to URL validation warnings

  #334·GSCSD1 opened Jun 30, 2026Contributor·

  ·
  1

• convert <img> tag text content to image_url for VLM models

  #311·vidushig-nv opened May 28, 2026Collaborator·

  ·
  1