Pika H3 Flex generates 15s video in 7 seconds, 80% cheaper than H3Max
HeyNayeem · x · 2026-11-06
Pika launched H3 Flex, a post-trained video model that generates faster than real time: a 15-second clip with soundtrack can be produced in as little as 7 seconds at 768p, averaging 1.8x faster than H3Max and up to 80% cheaper.
- Pika claims it's the #1 open-weight model on DesignArena's text-to-video board
- Supports resolutions up to 4K, up to 12 reference inputs, and precision editing via references and prompts
- Introduces exportable audio stems for isolating and refining any layer of the soundscape
- Designed for professional iteration: draft at 768p, then refine
User tests show a single prompt handling multi-shot storytelling, complex camera moves, and scene continuity. The model and its API are live now.
More from Multimodal
- SenseTime launches SenseNova 6.8 Flash with long-term memory, Blender and animated PPT output — heyshrutimishra · 2026-11-06
- Garden Skills' motion-showreel turns Claude Code into a code-rendered video studio — thisdudelikesAI · 2026-11-06
- Funny AI-animated short film "Morrigane and Nox" showcases narrative video generation — Outside-Objective828 · 2026-11-06
- AI oil paintings: wizard bird, long-nosed shades guy and a flame-snorting cactus cowboy — OVolosin82152 · 2026-11-06
- Viral prompt turns image models into character concept artists with ink-wash style sheets — azed_ai · 2026-11-06
- Screengrab reveals AI dubbing studio: sentence-level transcript editing with timed export — umesh_ai · 2026-11-06