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

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

News

• [2026/10] 🗺️ A unified runtime for interactive intelligence: native full-duplex models, turn-based multimodal models, and cascaded systems, coordinated through composable graphs and persistent sessions. [Blog]
• [2026/10] 🚀 SGLang-Omni v0.2.0 is on PyPI. Install with uv pip install --prerelease=allow "sglang-omni==0.2.0". [Installation]
• [2026/10] 🐟 Day-0 support for Fish Audio Drama 3: multi-speaker text-to-speech with voice cloning and natural-language control, streamed on /v1/audio/speech. [Blog] [Cookbook] [Technical Report]
• [2026/10] 🎙️ Introducing Voxt, an open-source voice input tool for Mac, now integrating SGLang-Omni for faster local speech recognition on Apple Silicon. Combine ASR with LLM-based text refinement and personal system prompts for accurate technical terminology and personalized voice input, with fully local or remote inference options. [Download][Demo]

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: Qwen3-Omni, MiniCPM-o 4.5, Ming-Omni, Ming-3.0-audio-flash — multimodal understanding and text/audio replies.
• Native full-duplex conversation: MiniCPM-o 4.5, PersonaPlex, NemotronLabs VoiceChat — live listening and speaking over /v1/realtime, including camera input with MiniCPM-o. Full-duplex guide.
• Cascaded voice assistants: Compose ASR → LLM → TTS services into streaming, turn-based conversations with voice activity detection, conversation history, turn taking, and interruption. Streaming cascade guide.
• Image and video generation: LLaDA2.0-Uni for text/image understanding, image generation, and editing; Cosmos3 Nano for interleaved understanding and generation, with native image, video, and action routes.
• Music generation: MiniMax Music 3 — lyrics + caption → 32 kHz stereo song.
• Speech generation: Higgs Audio v3, MOSS-TTS, MOSS-TTS Local, Fish Audio Drama 3, Fish Speech S2-Pro, Qwen3-TTS, Fun-CosyVoice3, Voxtral TTS, Ming-Omni-TTS, Ming-3.0-TTS, dots.tts, ZONOS2 — /v1/audio/speech, with batch, streaming, and voice features varying by model.
• Instruction-driven speech generation and editing: AuK and AuK-Flash — text and voice instructions → speech; audio and editing instructions → edited speech.
• Audio transcription: Qwen3-ASR, Fun-ASR, ARK-ASR, Nemotron 3.5 ASR, Nemotron Transcribe, Ming-3.0-ASR, experimental Whisper, and MOSS-Transcribe-Diarize via /v1/audio/transcriptions. MOSS-TD also returns speaker labels and timestamps (response_format=verbose_json).
• Speaker diarization: Nemotron 3 Diarization — speaker intervals for uploaded recordings and live audio, including overlapping speech.
• Speech translation: Hy-SimulMT — Chinese speech → English text and English speech → Chinese text, with file and realtime translation. See the audio translation guide for endpoint and model support.

Hardware Support

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

Quick Start

• Installation
• TTS usage
• ASR cookbook

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.

Citation

If you use SGLang-Omni for your research, please cite our technical blogs on the unified interactive runtime and the Stage--Process--GPU topology:

@misc{sglangomniinteractive2026,
  title={Unified Runtime for Full-Duplex Interactive Intelligence with {SGLang Omni}},
  author={{SGLang-Omni Team}},
  howpublished={\url{https://www.sglang.io/blog/omni-interactive-runtime}},
  year={2026},
  month={oct}
}

@misc{sglangomni2026,
  title={Fish Audio Drama 3 Meets {SGLang-Omni}: Day-0 Support and the Stage--Process--GPU Topology},
  author={{SGLang-Omni Team}},
  howpublished={\url{https://www.sglang.io/blog/spg-topology}},
  year={2026},
  month={oct}
}

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.