BrainbabyOur product
Brainbaby AI Video Engine
- How it works
- The Brainbaby AI Video Engine routes each prompt to the video model best suited to it, including ByteDance Seedance 2.5 and Alibaba Wan, and rewrites the prompt in the style that model responds to best.
- Max clip length
- Text-to-video: 2 to 15 seconds, or 20, 25 or 30 seconds. Image-to-video: 5, 10 or 15 seconds. In chat: 5 seconds.
- Resolutions
- Text-to-video: 480p, 720p or 1080p in landscape (16:9), portrait (9:16) or square (1:1). Image-to-video and editing: 720p or 1080p.
- Access
- Brainbaby web and mobile apps: Video Studio and chat, in the same account as chat, images, documents and presentations. No API keys.
- Price
- Video credits, priced by length and resolution (480p: 5 s = 1, 30 s = 4; 720p: 5 s = 2, 30 s = 7; 1080p: 5 s = 3, 30 s = 16 credits). The lowest-cost pack is 25 credits for $37, $1.48 a credit; packs start at 2 credits for $5. The studio shows the credit cost before you generate.
- Free tier
- No free video credits on the Free plan. Pro+ ($100/month) includes 7; Max ($200/month) includes 28 video credits a month.
- Best for
- Creators who want the best model for each idea without choosing one, comparing APIs or writing model-specific prompts.
Verdict: One prompt, the model best suited to it, and a prompt tuned for that model, in one workspace with chat, images and documents.
Sources: Brainbaby plans and credit packs
ByteDance
Seedance 2.5
- Max clip length
- 4 to 30 seconds in one pass. Dreamina adds a long-video beta of up to 180 seconds.
- Resolutions
- 480p, 720p or 1080p at 24 fps. Seedance 2.5 Flash: 480p or 720p.
- Native audio
- Yes. Audio and video are generated together, and audio clips can be used as references.
- Access
- Dreamina and CapCut (international), Jimeng and Doubao (China); API on BytePlus ModelArk, OpenRouter, fal, Replicate and Atlas Cloud; also in Freepik and Higgsfield.
- Price
- BytePlus: $10.70 per million video tokens, about $0.103 per second at 480p and $0.231 at 720p; 1080p is $11.70 per million, about $0.569 per second. Seedance 2.5 Flash: about $0.152 per second at 720p.
- Free tier
- Daily free credits in Dreamina and CapCut. We found no free Seedance 2.5 API trial on BytePlus.
- Best for
- Long single takes with sound that must reuse many reference images, clips or voices.
Verdict: Accepts more references than any other model here (up to 50), but it costs more per second than most rivals above 480p, and BytePlus is not available in the United States.
Sources: ByteDance Seed: Introducing Seedance 2.5, BytePlus: Dreamina Seedance 2.5 product page, BytePlus ModelArk pricing documentation, Dreamina: Seedance 2.5
Kuaishou
Kling 4.0
- Max clip length
- 3 to 30 seconds in one pass.
- Resolutions
- 720p and 1080p are priced in the API. Kuaishou has announced 4K.
- Native audio
- Yes, two-channel stereo.
- Access
- Kling web and mobile apps, the Kling API and resellers such as fal.
- Price
- The Kling API bills in units: 1.0 unit per second at 720p and 1.5 at 1080p, with the unit price set by the package you buy. fal lists $0.157 per second at 720p and $0.235 at 1080p.
- Free tier
- Free credits for new accounts; the amount is not published.
- Best for
- Longer clips with stereo sound and keyframe control (up to 10 keyframes and 15 references, per Kuaishou).
Verdict: A strong spec sheet for 30-second clips with stereo audio. The official API price is in units, so compare a real quote before you commit.
Sources: Kuaishou: Kling 4.0 launch announcement, Kling: Kling 4.0 model release (developer), Kling API pricing, fal: Kling 4.0 text-to-video, Kling AI credits policy
Google
Gemini Omni 1.2 Flash and Veo 3.1
- Max clip length
- Gemini Omni 1.2 Flash: 15 seconds. Veo 3.1: 4, 6 or 8 seconds.
- Resolutions
- Omni 1.2 Flash: 360p, 720p, 1080p or 4K. Veo 3.1 Lite: 720p or 1080p.
- Native audio
- Yes, for both. Omni also accepts audio as an input.
- Access
- Gemini app and Google Flow on Google AI plans, Google Vids, YouTube Shorts Remix and YouTube Create; Gemini API and Google Cloud. In the Gemini app, Omni replaces Veo 3.1.
- Price
- Gemini Omni 1.2 Flash: $0.07 per second at 720p. Veo 3.1 Lite: $0.05 per second at 720p and $0.08 at 1080p, with audio.
- Free tier
- Google Vids: up to three 15-second Omni videos a month at no cost. Gemini Omni Flash in YouTube Shorts Remix and YouTube Create at no cost (18+).
- Best for
- Short clips with sound at a low price, and free vertical clips for YouTube Shorts.
Verdict: Low prices and free ways to try short clips with sound. The 15-second ceiling means long takes have to be stitched together.
Sources: Google: Gemini Omni 1.2 Flash, Google Cloud: Gemini Omni 1.2 Flash documentation, Google: Introducing Gemini Omni, Google Workspace: Omni 1.2 Flash in Google Vids, Gemini Developer API pricing (Veo 3.1 Lite)
Runway
Runway Gen-4.5
- Max clip length
- 2 to 10 seconds.
- Resolutions
- 720p.
- Native audio
- No audio track is listed for Gen-4.5.
- Access
- Runway web app and the Runway API.
- Price
- API: 12 credits per second at $0.01 a credit, so $0.12 per second. In the app, the Standard plan is $12 a month billed annually ($15 monthly) for 625 credits.
- Free tier
- 125 one-time credits, about 10 seconds of Gen-4.5.
- Best for
- Teams that also want Runway’s editing tools in the same app.
Verdict: A clear per-second price, but 10 seconds at 720p without audio is a short spec sheet in late 2026.
Sources: Runway API pricing, Runway: Creating with Gen-4.5, Runway plans and pricing
Pika
Pika H3 Flex
- Max clip length
- 5 to 15 seconds.
- Resolutions
- Generated at 480p or 768p; Pika’s upsampler can finish a clip at 1080p, 2K or 4K.
- Native audio
- Yes, with optional separate dialogue, music and effects stems as .wav files.
- Access
- Pika web app, iOS app and the Pika API.
- Price
- API text-to-video: $0.01 per second at 480p and $0.02 at 768p. In the app, Pika bills credits at $0.01 each.
- Free tier
- The Free plan has no monthly credits; credit packs can be bought without a subscription.
- Best for
- Cheap drafts and social clips where separate audio stems make editing easier.
Verdict: The lowest per-second API price in this guide, but native output is below 1080p, and upscaling costs extra.
Sources: Pika H3 Flex, Pika API pricing, Pika plans and pricing
MiniMax
MiniMax H3 (Hailuo)
- Max clip length
- Up to 15 seconds.
- Resolutions
- H3: 768p or 2K. H3-Max: 480p or 768p.
- Native audio
- H3 generates native stereo audio.
- Access
- Hailuo AI app, the MiniMax API and open weights under the MiniMax H3 Community License.
- Price
- MiniMax API: H3 is $0.08 per second at 768p and $0.13 at 2K; H3-Max is $0.05 per second at 480p.
- Free tier
- Hailuo advertises free credits for new users; the amount is not published.
- Best for
- Teams that want to self-host, or need 2K output.
Verdict: Open weights and a low API price make it a flexible budget pick. Clips stop at 15 seconds.
Sources: MiniMax H3 announcement, MiniMax API video pricing
Alibaba
Wan 3.0
- Max clip length
- 2 to 30 seconds in one pass.
- Resolutions
- 480p, 720p or 1080p.
- Native audio
- Yes. Up to 5 audio clips can be used as references.
- Access
- Alibaba Cloud Model Studio API and OpenRouter. A faster Wan 3.0 Prime variant costs more.
- Price
- Model Studio (international): $0.05 per second at 480p, $0.10 at 720p and $0.20 at 1080p. Alibaba marks this as a limited-time discount, so check the console.
- Free tier
- A free quota of 30 seconds of video on Model Studio, valid for 90 days.
- Best for
- 30-second clips with sound at a low price per second.
Verdict: The cheapest way we found to get a 30-second 1080p clip with sound in one pass.
Sources: Alibaba Cloud: Wan3.0 officially launched, Alibaba Cloud Model Studio: wan3.0-video, Alibaba Cloud Model Studio model pricing, OpenRouter video model catalog (prices, durations, resolutions)
Luma AI
Luma Ray3.2
- Max clip length
- Up to 20 seconds.
- Resolutions
- 360p (draft), 540p, 720p and 1080p, with HDR options.
- Native audio
- Luma does not document audio for Ray3.2.
- Access
- Luma web app and the Luma API.
- Price
- 100 credits for 5 seconds at 720p and 400 at 1080p. Luma estimates this at about $0.06 per second at 720p and $0.24 at 1080p.
- Free tier
- Not published.
- Best for
- Shots that need keyframe control (up to 16 keyframes) or HDR output.
Verdict: Good control features, but prices are in credits and audio is not documented.
Sources: Luma: Introducing Ray3.2, Luma plans and pricing, Luma: AI video cost per second
xAI
Grok Imagine Video 1.5
- Max clip length
- 1 to 15 seconds.
- Resolutions
- 480p, 720p or 1080p.
- Native audio
- Yes.
- Access
- xAI API and OpenRouter.
- Price
- Grok Imagine Video 1.5: $0.08 per second at 480p, $0.14 at 720p and $0.25 at 1080p. 1.5 Lite: $0.02, $0.03 and $0.14.
- Free tier
- We could not verify a free tier.
- Best for
- Cheap 720p drafts with sound (Lite), or references and voices (full model).
Verdict: Lite is one of the cheapest 720p-with-audio APIs we found; switch to the full model for references and native 1080p.
Sources: xAI docs: Grok Imagine Video 1.5, xAI docs: Grok Imagine Video 1.5 Lite, OpenRouter video model catalog (prices, durations, resolutions)
Black Forest Labs
FLUX.3 Video
- Max clip length
- 5 to 20 seconds, as listed on OpenRouter.
- Resolutions
- 720p, 1080p, 2K and 4K.
- Native audio
- Optional, including dialogue and sound effects, at no extra charge.
- Access
- Black Forest Labs API and OpenRouter.
- Price
- Drafts from $0.06 per second. Full renders start at $0.15 per second at 720p on BFL’s pricing page; its documentation lists $0.17, so check your quote.
- Free tier
- Not published.
- Best for
- Draft-then-finish workflows that end in a 4K render.
Verdict: Cheap drafts and a path to 4K finals; BFL’s own rate cards disagree, so confirm the price.
Sources: Black Forest Labs: FLUX.3 Video, Black Forest Labs API pricing, OpenRouter video model catalog (prices, durations, resolutions)
OpenAIDiscontinued
Sora 2
- Status
- OpenAI shut down the Sora web and app experiences on April 26, 2026 and the Sora API on September 24, 2026.
- Best for
- Nothing new: Sora can no longer be bought or used.
Verdict: Do not buy a plan or tool expecting Sora. Choose a current model instead.
Sources: OpenAI: What to know about the Sora discontinuation