fal dev spills: MiniMax H3 'quality' mode matches balanced in 99.9% of cases but adds 60-120s
isidentical · x · 2026-09-04
fal developer isidentical revealed internals of MiniMax H3 Max Turbo's 'prompt expansion mode': the quality setting invokes MiniMax's proprietary prompt expansion model, taking over 60-120 seconds, yet in their evals quality performs on par with the default balanced mode in 99.9% of cases. He suggested removing the option entirely since 'so many people fall for it,' and recommends balanced for actual speed.
This responds to altryne's live test where Turbo generated slower than Max with poor prompt adherence. The post includes fal's H3 Max Turbo API page and a sample expanded prompt output.
Related event: MiniMax H3 Max Turbo tested slower than Max; quality mode called useless(3 posts)→
More from Multimodal
- Runway joins Claude Marketplace as its first creative partner — tlakomy · 2026-10-21
- Claude Code directed a full music video: 6.5 hours, 765M input tokens — colinmcnamara · 2026-10-21
- Kling 4.0 debuts native 30-second AI video with 10 keyframes, extendable to 2 minutes — lmoroney · 2026-10-21
- Mistral's Voxtral transcribes an 1891 Eiffel phonograph recording at 6.5% WER — IsForAt · 2026-10-21
- Redditor remakes SD clown dance video at 25 steps with motion boost and fixed prompts — roychodraws · 2026-10-21
- Fibo-2, the #1 open image editing model, releases weights on Hugging Face — linoy_tsaban · 2026-10-21