vLLM-Omni v0.22 adds Cosmos 3, robot serving, and broader quantization support
MonaJalal_ · x · 2026-07-22
vLLM-Omni v0.22.0 is a major upgrade focused on production-grade multimodal serving.
It adds day-0 support for NVIDIA Cosmos 3 world models, robot serving via DreamZero + OpenPI realtime API, production TTS for models like Qwen3-TTS and Qwen3-Omni, faster image/video/diffusion support including Wan 2.2, HunyuanVideo 1.5, and LTX-2.3, plus broader quantization and hardware coverage such as FP8/INT8, MXFP4/MXFP8, and W4A16. The release notes mention 339 commits, 124 contributors, and 52 new contributors.
More from Infra
- Oracle Q1: Triple-digit cloud infrastructure growth as AI demand booms — DavidLinthicum · 2026-09-11
- SF Compute founder: buying compute is 'an absolutely awful experience' right now — IgorCarron · 2026-09-11
- SmolVM open-sources persistent computer infrastructure for agents that outlive chat sessions — aniketmaurya · 2026-09-11
- PyTorch Day Korea 2026 launches first offline conf, CFP closes Sept 13 — PyTorch · 2026-09-11
- Local LLM server dilemma: 4x CMP-170HX (price up 53% in 20 days) vs Mac Studio M5 Ultra — rumboll · 2026-09-11
- llama.cpp lands Flash Attention tuning for RDNA4, big prefill gains on AMD — pmttyji · 2026-09-11