Yang Gao

Member of Technical Staff @MosiAI_Official | ex-Shanghai AI Lab Pretrain Team @intern_lm | Dev @OpenMMLab #MMRazor | Building open-source MLLM & ML systems

[Yang Gao @YangGao07]

Everything is a plugin👍

[DeepSeek @deepseek_ai]

🧩 DeepSeek Harness v0.1 is now available in Developer Preview!

🔹 We’re opening it up to developers building agent harnesses worldwide and open-sourcing the codebase in MIT license.
🔹 Powered by the Cordis meta-framework, DeepSeek Harness is an agent harness built around one core idea: Everything is a plugin. Models, tools, skills, sessions, sandboxes, filesystems, loops, orchestration, and UI are ALL implemented as plugins, and can be mixed, matched, replaced, and extended.

Try it now!
https://t.co/2YWSvJHhKA

[Yang Gao @YangGao07]

@JeffDean Legend!

[Yang Gao @YangGao07]

Even in the vibe-coding era, the SGLang-Omni team continues to set the bar for rigorous systems engineering—solid profiling, careful optimization, and results that speak for themselves. Seriously impressive work. Huge kudos to the team! 🚀

[Yichi Zhang @YichiZ03]

The hardest part of maintaining an open-source project is serving them at scale while keep high code quality.

In SGLang-Omni, we spent a month refactoring, removing 2000+ non-test duplication. Here is what we changed, and what we learned. https://t.co/kaxYsYokmo

[SGLang @sgl_project]

Count us in! SGLang community has co-signed the Open Weights letter @lmsysorg

Everything we ship goes out in the open, because world-class performance should be accessible to every builder.

Let's keep building in the open 🧡

[Sonya Huang 🐥 @sonyatweetybird]

The first Vera Rubin clusters are here!

Yesterday, @IneffableLabs took delivery of their Vera Rubin NVL72 cluster from @googlecloud @nvidia

The AI frontier jumps forward by yet another generation of hardware. Acceleration continues.

[Yang Gao @YangGao07]

Open source unites the world. 🌍🤝✨

[Anthony Pompliano 🌪 @APompliano]

Banning Chinese models is the most un-American thing we could do.

Stop complaining and start competing.

People won't use Chinese models if American companies build something better.

[Yichi Zhang @YichiZ03]

https://t.co/SbM1XD8dAf

[Yang Gao @YangGao07]

Solid work!

[Ratish P @ratishp22]

When the SGLang Omni team profiled MOSS-TTS Local v1.5, we expected autoregressive decoding to dominate. Instead, the largest bottleneck was its transformer vocoder. I built a packed FlashAttention path that increased QPS by 48.64% and reduced mean latency by 32.71%.

[Yang Gao @YangGao07]

@ShahzaibAli3029 @MosiAI_Official @huggingface We welcome community contributions to bring our models to third-party frameworks like CrispASR. We’ll open a GitHub RFC soon, with our team actively supporting the effort. Stay tuned!

[MOSI @MosiAI_Official]

🤗 MOSS-Transcribe-Diarize-0.9B is now open source on @huggingface.

Built with an end-to-end audio-to-structured-transcript paradigm:

>0.9B open-source ASR model
>Apache license 2.0
>128k long-context transcription
>Up to ~90-min audio input
>Speaker labels + timestamps in one generation
>Multi-speaker diarization for meetings, interruptions, and overlapping voices
>Hotword biasing for names, terms, and domain-specific vocabulary
>~100 token/s on NVIDIA RTX 4090, RTF ~0.017

Thank you @sgl_project @vllm_project @Prince_Canuma @lllucas for day-0 support! 🚀

Github: https://t.co/0hgIjvhGv1
Huggingface: https://t.co/gQ1avC3w2e
API: https://t.co/znzfO4PHuu
Live demo: https://t.co/wAmEbbFna5
Technical Report：https://t.co/3yb0VKp7FS
HF Space: https://t.co/gQ1avC3w2e
AtomGit：https://t.co/uIwqnkMeGU
SGLang-Omni: https://t.co/TFlp4YzI7f
vLLM: https://t.co/j1VMVuQDcp
MLX-audio: https://t.co/ofDO9V6N7J
Discord:https://t.co/H3shzNLHOL

[Yang Gao @YangGao07]

@GenAI_is_real 🔥🔥🔥

[Yang Gao @YangGao07]

@YichiZ03 Solid!

[Yang Gao @YangGao07]

Impressive work from SGLang-Omni Team, well done @YichiZ03 !

[Yichi Zhang @YichiZ03]

https://t.co/o9aPp3UjK1

[Yang Gao @YangGao07]

@ZDi____ 🔥🔥🔥

[Novita AI @novita_labs]

🚀 MOSS-TTS Local Transformer v1.5 is now available on Novita.

Build voice agents with:
🎙️ Zero-shot voice cloning
🌍 Speech synthesis in 30+ languages
🎧 Native 48 kHz stereo audio
⚡ Streaming TTS with ultra-low latency

Thanks to @MosiAI_Official for open-sourcing MOSS-TTS.

[Yang Gao @YangGao07]

Yes sir!!🫡🫡🫡

[Ivan Fioravanti ᯅ @ivanfioravanti]

MLX news: MOSS-TTS-Local Transformer 1.5 available on mlx-audio now! Thanks to @Prince_Canuma and @lllucas

Audio on! 🔉

[Yang Gao @YangGao07]

smooth👏👏👏

[Yang Gao @YangGao07]

@3___infinix @ivanfioravanti @MosiAI_Official 您好，local transformer 架构的 TTFT 在设计上会有显著优势，我们本地测试最低可达 <100ms，欢迎在您的硬件上测试反馈。此外，我们开放了 API 平台：https://t.co/Z6k0hiU8m7

[Yang Gao @YangGao07]

More tech blogs on the way🔥🔥🔥

[LMSYS Org @lmsysorg]

Check out the latest blog on how SGLang Omni supports MOSS-TTS-Local-Transformer-v1.5: https://t.co/pFCBvOLP6Y

[Yang Gao @YangGao07]

We’re glad to see the community building around MOSS-TTS Local Transformer v1.5.

A developer has created a ComfyUI custom node for MOSS-TTS, helping bring local text-to-speech generation into ComfyUI workflows.

Thanks to the contributor — 🌟https://t.co/Bjio3vtSMD