MiniMax pricing splits text, motion and sound
See how MiniMax’s M-series token pricing relates to separate H3 video endpoint rates and why a single quote can need two units.

👁 111 reads
A promotional clip can begin with an M-series model writing a script and end with an H3 endpoint rendering motion. MiniMax lists M3.1 Flash at $0.15 per million input tokens and $0.60 per million output tokens, with cache reads at $0.015 per million. Those figures cover text processing.
For H3 Max video, fal’s standard rates from November 16 are $0.05 per second at 480p, $0.08 at 768p and $0.16 at 1080p. A five-second 768p generation at the listed standard rate comes to $0.40. The October and early November promotions appear on the same page; their end dates matter.
I would keep the script constant and audition the motion at the intended resolution before the editor books a revision round. On a small US studio job, repeating the clip can outweigh the entire text bill.
Practical details
The fal H3 Max page switches back to standard rates from November 16, with separate rates by output resolution.
Google · Your sourcesQuantslant higher in your Google - one clickAdd to preferred sources →Sources
- https://platform.minimax.io/docs/guides/pricing-paygo (accessed November 17, 2026)
- https://fal.ai/models/minimax/h3-max/text-to-video (accessed November 17, 2026)
More on this from QUANTSLANT
ai tools·Microsoft MAI models for images, speech and codeChoose a Microsoft MAI endpoint by deliverable, from a targeted product edit through transcription to a reviewed software patch.