I gave Nemotron 3.5 Lightning eyes and ears with zero training, because the architecture let me

NVIDIA's Nemotron 3.5 Lightning, a 30B-A3B hybrid Mamba MoE agent model released two weeks ago, has been extended to process images and speech without any training, thanks to a geometry coincidence with NVIDIA's Nemotron-3-Nano-Omni. The extension, called Lightning-Omni, attaches the Omni's projectors to Lightning zero-shot, enabling image and audio understanding on stock llama.cpp, with video support via a fork. This demonstrates that perception can be a detachable organ within an architecture family.

NVIDIA released Nemotron 3.5 Lightning two weeks ago: a 30B-A3B hybrid Mamba MoE agent model with 1M context, text only. Tonight I released Lightning-Omni: the same model reading images and understanding speech, with no training involved. The trick is a geometry coincidence that is not really a coincidence. Lightning shares the exact backbone shape of NVIDIA’s Nemotron-3-Nano-Omni nemotron h, hidden 2688, 52 layers, same 131072 vocabulary . The Omni’s projectors were trained by NVIDIA to translate C-RADIO vision features and Parakeet audio features into precisely that embedding geometry. So the towers and projector attach to Lightning zero-shot: download, load, works. What I measured before claiming anything: Every quantization passes text, image, and audio gates before upload; the graded matrix ships in the repository and grows as tiers land: Runtime: image, text and audio run on stock llama.cpp with these files. Video needs my fork the Nemotron video graph is not upstream yet , with prebuilt arm64 CUDA binaries: This is also a small existence proof for a general pattern: within an architecture family, perception can be a detachable organ. Any future nemotron h checkpoint with this geometry can be given the same eyes and ears in an evening. Questions welcome, and if you try it on your own hardware I would love to see results.