Research3 min read
TII's Falcon Perception Beats SAM 3 on Visual Grounding — With a 0.6B Model That Runs on One GPU
The Technology Innovation Institute has released Falcon Perception, a 0.6-billion-parameter early-fusion Transformer that outperforms Meta's SAM 3 on open-vocabulary visual grounding while running on a single GPU. The model introduces PBench — a diagnostic benchmark that separates perception capabilities by complexity — and ships alongside Falcon OCR, which achieves the highest throughput of any open-source OCR model.
D.O.T.S AI Newsroom3 min read