Live aus den gerenderten CRDs (LLMInferenceService, Pods, Ingresses)
in vllm auf amd-ai. JSON: /api/lanes.
Kundenzugang: api.ai.x-ion.org (Aliase siehe Modellkatalog).
Stand: 2026-08-18T15:29:36Z
| Lane | Modell | Ready | TP | GPUs | Ingress | Interner Gateway-Pfad | Image |
|---|---|---|---|---|---|---|---|
asr-whisper | openai/whisper-large-v3 | True | 1 | 1 | asr.xlab-ha.devops.x-ion.org | http://kserve-ingress-gateway.kserve.svc.cluster.local/vllm/asr-whisper | llm-d-router-endpoint-picker:v0.9.0 |
image-flux2-klein | flux2-klein | True | 1 | 1 | image.xlab-ha.devops.x-ion.org | http://kserve-ingress-gateway.kserve.svc.cluster.local/vllm/image-flux2-klein | llm-d-router-endpoint-picker:v0.9.0 |
tts-qwen3 | Qwen/Qwen3-TTS-12Hz-1.7B-CustomVoice | True | 1 | 1 | tts.xlab-ha.devops.x-ion.org | http://kserve-ingress-gateway.kserve.svc.cluster.local/vllm/tts-qwen3 | llm-d-router-endpoint-picker:v0.9.0 |
Zukunft (XXL-60): geplante Trainings, Reservierungen, Mutation per LLM-Query/CRD.