62 1 week ago

Context-optimized variant of Google's Gemma 4 12B model, tuned for GPU-constrained inference.

vision tools thinking