H3 Max benchmark: 15s 768p video generated in 17s on fal
umesh_ai · x · 2026-08-27
Tested the H3 Max video model on the @fal platform. Generating a 15-second video at 768p resolution took only about 17 seconds and cost $0.60. This performance is effectively real-time by current video model standards.
Related event: H3 Max Tested: 15-Second 768p Video Generated in as Fast as 5 Seconds(2 posts)→
More from Multimodal
- SenseNova U1.5-Lite Test Results: Style Transfer and Local Editing Demos — PrajwalTomar_ · 2026-08-27
- SenseNova U1.5-Lite: 8B Open-Source Model Enables Precise Local Image Editing — PrajwalTomar_ · 2026-08-27
- Seedance 2.5 Workflow: Create Anime Stories with Consistent AI Characters — aftahi_ai · 2026-08-27
- SDXL+LoRA struggles to match Nano Banana for commercial watercolor style transfer — BroadPomegranate5652 · 2026-08-27
- Seedance 2.5 Launches: Generate 30-Second AI Dance Videos — aftahi_ai · 2026-08-27
- Red Umbrella in Flooded Library: Midjourney Context Prompting — tisch_eins · 2026-08-27