αalphaBell / MiniMax Video Budget
Source-pinned model lens

One video. Many visual slots.

See how sampled frames become vision placeholders in MiniMax M3.1 Flash.

16 sampled frames · 672 × 672 processed pixels
8 groupstwo frames each
×
576placeholders per group
=
Vision placeholders
4,608

From the pinned public processor

Processor arithmetic. Excludes text, timestamps, memory and speed.Interactive inspector ↗