LFM2.5 VL 3B DSpark released

Pairing LFM2.5-VL-3B with this small 280M-parameter draft model cuts decode latency substantially on both H100 and Apple silicon, a practical lever for anyone serving vision-language models without touching output quality.

• articleAccelerating vision-language models with LFM2.5-VL-DSparkhuggingface.co
• articleUp to 3.2x Faster Inference with LFM2.5-DSparkhuggingface.co
• postDSpark draft models cut LFM2.5 agent latency roughly in halfLiquid AI

Checking sign-in…

Loading comments…