sglang-omni 0.1.1

Blog | Documentation | Quick Start | Cookbook | SGLang | Join Slack

⭐ Star SGLang-Omni to help more builders discover open infrastructure for multimodal and speech serving!

News

• [2026/08] 🚀 SGLang-Omni completed its TTS architecture refactor, consolidating pipeline state, engine construction, reference encoding, capability metadata, and streaming and non-streaming vocoder scheduling into reusable framework components. [Roadmap] [Blog]
• [2026/06] 🔥 MOSS-TTS Local Transformer v1.5 runs on SGLang-Omni with native-streaming 48 kHz speech. [Blog] [Cookbook]
• [2026/06] 🔥 Higgs Audio v3 TTS runs on SGLang-Omni for real-time, controllable speech for voice agents. [Blog] [Cookbook]

About

SGLang-Omni is a multi-stage serving runtime for omni, speech, and TTS models. Its design target is multi-stage decoding: generation split across heterogeneous stages with different compute patterns, dependency structures, and resource needs. SGLang-Omni owns the pipeline topology, stage lifecycle, inter-stage transport, model-family integration layer, and OpenAI-compatible serving surface, while composing with SGLang for high-performance autoregressive scheduling and model execution where applicable.

• Multi-stage runtime: SGLang-Omni models generation as coordinated stages: preprocessing, encoders, autoregressive engines, talkers, decoders, vocoders, and aggregators.
• Stage-specialized scheduling: Each stage runs behind a scheduler matched to its workload, from SGLang-backed autoregressive scheduling to lightweight preprocessing and streaming vocoder loops.
• Transport-aware execution: A control plane coordinates requests while the relay data plane moves tensor payloads across shared-memory, NCCL, NIXL, and Mooncake backends.
• API surface: OpenAI-compatible endpoints expose multimodal chat, speech generation, batch speech, streaming speech, uploaded voices, and transcription.

What SGLang-Omni Serves

• Omni chat and speech: Run models such as Qwen3-Omni and Ming-Omni with multimodal inputs, text/audio outputs, and thinker-talker generation pipelines.
• Speech generation: Serve Higgs Audio v3, MOSS-TTS, MOSS-TTS Local, Fish Speech S2-Pro, Qwen3-TTS, Voxtral TTS, Ming-Omni-TTS, and related TTS systems through speech, batch speech, streaming speech, and uploaded-voice APIs.
• Audio transcription and diarization: Serve Qwen3-ASR and MOSS-Transcribe-Diarize through the OpenAI-compatible /v1/audio/transcriptions endpoint. MOSS-TD supports speaker-labelled, timestamped transcript segments with response_format=verbose_json.
• SGLang-Omni Router: Serve multiple Omni servers behind one OpenAI-compatible endpoint, with health checks, readiness tracking, worker lifecycle control, and model-capability discovery across the worker pool. See the Router guide.

Additional model guides, including experimental and research-oriented paths, are available in the Cookbook.

Quick Start

• Installation
• TTS usage
• Qwen3-Omni usage
• Qwen3-ASR cookbook
• MOSS-Transcribe-Diarize cookbook
• Omni router
• Developer reference

Community & Support

SGLang-Omni welcomes contributors working on inference systems, kernels, scheduling, inter-stage communication, model runners and cache efficiency, model integration, benchmarking, production deployment. Join the SGLang Slack or read the developer reference.

Organizations interested in supporting SGLang-Omni, TTS, or omni model serving can contact Chenyang Zhao at zhaochenyang@lmsys.org.

Acknowledgments

SGLang-Omni builds on the SGLang ecosystem and on open model work from the TTS, speech, and omni-model communities. We thank the model teams, systems contributors, and partner organizations helping make open multimodal serving faster, more reliable, and easier to extend.