// news · multimodal · industry2026-05-23source: openai / verge / decoder

OpenAI launches Sora 2.5 — 90-second video at 4K, audio-synced lip-sync, $200/month consumer tier

OpenAI launched Sora 2.5 this week, the second major Sora generation since Sora 2 in December 2025. The 2.5 release pushes max clip length to 90 seconds at 4K resolution with audio-synced lip-sync as a first-class feature, and introduces a $200/month consumer tier alongside the existing $20 ChatGPT-Plus integration tier.

The 90-second-at-4K clip length is the production threshold that matters for short-form video workflows (TikTok, YouTube Shorts, Instagram Reels are all in the 60-90 second band). Sora 2.5 moves the consumer-tier output from 'impressive demo' to 'usable for short-form publish', which is the first time a hosted video model has hit that threshold without serious post-production cleanup.

The $200/month consumer tier is the structural pricing signal. OpenAI is testing whether premium video-generation is a separable consumer subscription line or a ChatGPT-Plus add-on. The two-tier launch suggests OpenAI thinks the premium price point holds for a subset of professional creators while keeping the entry point bundled into the existing ChatGPT subscriber base.

See our analysis →

OpenAI — Sora 2.5 launch announcement → · The Verge — Sora 2.5 ships at $200/month → · Decoder — What Sora 2.5 changes about short-form video →