Pika H3 Flex launches: faster-than-realtime video model, 1.8x quicker and 80% cheaper than H3Max
minchoi · x · 2026-10-30
Pika released H3 Flex, a post-trained H3 video model that generates faster than real time in nearly every configuration — a 15-second clip in as little as 7 seconds, on average 1.8x faster and 80% cheaper than H3Max. Built for professional iteration: up to 4K resolution, up to 12 reference inputs, draft-at-768p/finish-at-4K workflow, and exportable audio stems for isolating parts of the soundscape. Available via the Pika API.
More from Multimodal
- AI design accidentally triggers a human renaissance: taste beats execution by 2028 — EXM7777 · 2026-10-30
- Voice agents have a moment: Rime Prism, Deepgram buys Daily, Phonic tops GPT-Live — rdesh26 · 2026-10-30
- Fal's H3 Max still unreleased 2 months later as Pika's H3 Flex overtakes it on the leaderboard — TomLikesRobots · 2026-10-30
- MiniMax H3 emotional depth test turns a vague prompt into a tearful golden retriever — Ded_Ghost_007 · 2026-10-30
- PixVerse first-person door-opening video looks too real to dismiss — Ok-Line2658 · 2026-10-30
- Armin Ronacher on Qwen3-TTS: custom game narrator voices, cloning limits and workarounds — mitsuhiko · 2026-10-30