feat: MLLM prefix caching with 3x speedup#13

Closed

lubauss wants to merge 8 commits intowaybarrios:mainfrom

lubauss:patch/feat-mllm-prefix-caching-with-3x-speedup

Commits on Jan 19, 2026

feat: Enable continuous batching for MLLM models (#1)

lubauss
and
claude
authored
fix: Enable vision and streaming for MLLM models + Gemma 3 support (#2)

lubauss
and
claude
authored
fix: Handle Gemma 3 required pixel_values parameter in continuous batching (#4)

lubauss
and
claude
authored