Gemini Omni Flash 1.2 takes #2 on video editing, cuts price 31% to $6.26/min
ArtificialAnlys · x · 2026-11-11
Artificial Analysis published its full evaluation of Google's latest video model Gemini Omni Flash 1.2 (released Nov 5):
- Rankings: #2 on AA-Video-Editing v1.1, up from #12 to #6 on AA-Video-T2V v2.1 and #17 to #14 on AA-Video-I2V v2.1. Supports text-to-video, image-to-video and video editing with native audio, clips up to 15 seconds; replaces 1.1 on the Gemini API on Dec 3.
- Strengths: dialogue scenes (#1 Dialogue & Lip Sync), multi-shot narrative (#2), UI/motion design (#1); #1 in both Complex Edits and Visual Effects Editing, ahead of Wan 3.0.
- Weaknesses: live-action realism, human anatomy (#9/#12), sound & speech editing (#8), physics & world simulation slipping to #6.
- Speed: a 10-second 1080p clip in 76 seconds, faster than every higher-ranked model (Wan 3.0 and Dreamina Seedance 2.5 take 4 minutes).
- Price: $6.26/min for 1080p, 31% cheaper than 1.1 ($9.12); MiniMax H3 ($4.80) and HeyGen Video 1 ($2.40) remain cheaper.
- Available via Gemini API, Google AI Studio, Vertex AI, the Gemini app, Flow and YouTube.
More from Multimodal
- fal Agent lands in Photoshop, bringing multi-model image and video generation in-app — gorkem · 2026-11-11
- Claude writes all code from one photo: photorealistic sunlit room rendered in vork — ravisparikh · 2026-11-11
- Clay octopus siege test: none of four image models nails eight tentacles — aziz4ai · 2026-11-11
- Anime key-visual test: Nano Banana steadiest, GPT Image grandest but photoreal — aziz4ai · 2026-11-11
- 4 Image Models, 5 Art Styles: Hands-on Comparison of MiniMax H3, Seedream 5.5, Nano Banana 2.1 and GPT Image 2.6 — aziz4ai · 2026-11-11
- AI-generated 1972 retro Avengers is a delightfully weird demo — techhalla · 2026-11-11