Skip to content

vllm_omni.entrypoints

vLLM-Omni entrypoints module.

Modules:

Name Description
async_omni

AsyncOmni - Refactored async orchestrator using AsyncOmniEngine.

async_omni_base

AsyncOmniBase: the asyncio-side foundation shared by AsyncOmni and DuplexOmni.

cli

CLI helpers for vLLM-Omni entrypoints.

client_request_state
duplex

Thin websocket serving for engine-resident duplex sessions.

duplex_omni

DuplexOmni: the Python API for full-duplex models.

omni
omni_base
openai

OpenAI-compatible API entrypoints for vLLM-Omni.

openpi
pd_utils

PD (Prefill-Decode) disaggregation helpers.

serve
stage_utils
utils