Which AI is best for generating video?

In blind comparisons, Wan 3.0 from Alibaba currently leads text-to-video generation, with an Elo of 1175 on the Artificial Analysis leaderboard (read 30 October 2026). The gap is small, so Utopai X (1166) and MiniMax H3 (1161) are close alternatives. Seedance 2.5 and FLUX 3 Max both score 1157, while Google's best entry, Gemini Omni Flash 1.1, ranks 10th with 1128. OpenAI's Sora is no longer available: the app closed on 26 April 2026 and the API on 24 September 2026.

• Wan 3.0 (Alibaba): the top-rated model in blind comparisons.
• MiniMax H3: the model behind Hailuo, with clips up to 15 seconds, 2K resolution by default and native stereo sound.
• Seedance 2.5 (ByteDance): clips up to 30 seconds, rolling out in Dreamina and CapCut.
• Kling 4.0 (Kuaishou): launched in October 2026, generates up to 30 seconds in one pass and is also available inside Runway and Luma.
• Cost: Runway has a free plan with 125 one-time credits and a Standard plan at 15 USD per month (12 USD with annual billing). Luma lists no free plan, with Plus from 30 USD per month. Gemini Omni Flash in the Gemini API costs about 0.10 USD per second of 720p video.

Rankings change every few weeks, so don't rely on a single leaderboard. For short ads and social clips, run the same prompt and reference image through two models and compare the results. Check each platform's licence terms before using the output commercially.