MiniMax H3 on Mac: M3 Ultra Takes 2.6 Hours for a 10-Second Video
cocktailpeanut · x · 2026-08-06
A developer tested the MiniMax H3 video generation model locally on an M3 Ultra Mac. Using the Phosphene workflow (1280x768 resolution, 20 steps), generating a 10-second video took 2 hours and 39 minutes.
While currently slow, the developer noted there is room for optimization. Future tests are planned on the M5 Max, which has fewer cores (40 vs. 80) but utilizes a new architecture, to compare local inference performance across different hardware setups.
Related event: MiniMax H3 Local Inference Benchmarks: Runs on Consumer GPUs and Mac(16 posts)→
More from Infra
- AMD Acquires AI Chip Startup Taalas to Boost Model-Specific Accelerators — karlfreund · 2026-08-07
- AMD Acquires AI Chip Startup Taalas to Boost Decode Acceleration — BenBajarin · 2026-08-07
- Breaking the AI Memory Wall: CXL Moves to Deployment, Marvell Well Positioned — BenBajarin · 2026-08-07
- Samsung, SK Hynix, and Micron Sell Out 2027 Memory Capacity to AI Firms — ramos_casals · 2026-08-07
- DSpark Boosts Local DeepSeek-V4-Flash Inference Speed by 2x — petrusenko_max · 2026-08-07
- Nativ Launches Open-Source Local AI App for Mac, Prioritizing Privacy and Native Frontier Models — Scobleizer · 2026-08-07