Run LFM3-Omni-2B locally in the browser

Run the LFM3-Omni-2B omni-modal model 100% locally in your browser using custom WebGPU kernels. It eliminates the need for separate ASR or TTS models, handling visual, audio, or text inputs and generating interleaved text and audio tokens.

Reddit r/LocalLLaMA/u/xenovatech

LFM3-Omni-2B-WebGPUWebGPUHugging Face

Added 2026-10-31 · originally published 2026-10-30

Steps

1Open the LFM3-Omni-2B-WebGPU demo page on Hugging Face.
2Set up a WebGPU-compatible browser environment (using hardware like M4 Max for speeds around 280 tok/s).
3Input visual, audio, or text content to run end-to-end omni-modal interactions.

Copy-ready prompt

Drawing on the feature of the LFM3-Omni-2B model enabling local WebGPU omni-modal execution in the browser, help me analyze how to convert [Input type, e.g., visual/audio/text] into interleaved [Output type, e.g., text and audio] tokens.

View original post

#AI Models#WebGPU#Omni Modal#Browser Execution

Running LFM3 offline on mobile with ExecuTorch export

Export LFM3-1.7B and LFM3-VL-2B models to ExecuTorch to run them completely offline on iPhone and Android. Tests show LFM3-VL-2B can read a Japanese ramen menu offline and accurately answer questions about spice levels and payment methods.

Develop a browser-based PS2 emulator with Claude

A developer used Claude (Opus 5.5) to write core code, building the TabStationToo PS2 emulator directly inside browser tabs using the PCSX2 core alongside WebAssembly and WebGPU technologies. The project successfully implements local file reading, code transpilation, and automated benchmarking.

Found this useful? Get daily updates on Telegram or X.

Join on Telegram Follow @CaiXiang23766

Summarized by AI from the original source - refer to the original post for details. Copyright belongs to the original author.

Steps

Related use cases