asepharyana bcb563ea7f feat(vision): route multimodal analysis to dedicated NVIDIA direct endpoint ...

- config: add AI_LLM_VISION_BASE_URL + AI_LLM_VISION_API_KEY (separate from text router)
- llmClient: llmVision() now calls dedicated vision endpoint when configured
  (axios POST to integrate.api.nvidia.com, model nvidia/nemotron-3-nano-omni-30b-a3b-reasoning,
  reasoning_budget 16384, non-stream), falls back to router combo otherwise
- keeps text/moderation on omniroute, vision on NVIDIA direct

2026-08-15 14:31:53 +07:00

..

backend

fix(goLive): retry VIDEO(op12)/SPEAKING(op5) opcodes until ws OPEN — broken shared-screen video

2026-08-13 01:44:40 +07:00

discord-gateway

feat(vision): route multimodal analysis to dedicated NVIDIA direct endpoint

2026-08-15 14:31:53 +07:00

frontend

fix(voice): separate Mic and Listen state (were both bound to listen)

2026-08-14 12:35:44 +07:00