Models
Every top model, with your brand built in.
Creative Studio offers 13 video and 10 image models. Studio picks the right one for each request, or you choose. Everything you make can be scheduled to your social accounts from the same place.
Video models.
Kling 3.0 ProKuaishouCinematic motion with strong image-to-video support. Weaker at dialogue in Spanish and at logo fidelity.Kling 3.0 4KKuaishouNative 4K Kling output for sharper social exports.Seedance 2.5ByteDanceByteDance long-form video with native audio, 4–30s clips, and multimodal references.Seedance 2.0 MiniByteDanceFaster lower-cost ByteDance video model for quick social iterations.Grok Imagine 1.5xAILow-cost xAI video with native audio, 1080p output, and up to 7 reference images.FLUX 3Black Forest LabsBlack Forest Labs cinematic video with native audio at 720p or 1080p. New — Spanish speech and logo fidelity are unreviewed.Veo 3.1GoogleExpressive faces and emotional close-ups. In-video text and non-English dialogue can be weak.Gemini Omni Flash 1.1GoogleGoogle multimodal video with native synchronized audio, up to 4K, and an End frame on image-to-video.Happy Horse 1.1AlibabaAlibaba native-audio video model for dialogue, lip-sync, and reference-driven scenes.MiniMax H3 MaxMiniMaxfal's post-trained MiniMax H3 with native audio — faster generation with stronger prompt adherence at 480p, 768p, or 1080p. Auto's first choice for everyday video, Spanish included.MiniMax H3 Max TurboMiniMaxFaster, half-price MiniMax H3 Max with native audio at 480p, 768p, or 1080p. Text and start-image video only: no saved characters or reference images.Wan 3.0AlibabaAlibaba single-pass video with native audio, 2–30s clips, and multimodal references at up to 1080p. New — manual pick only until lane placement is evaluated.Kling 3.0 Turbo ProKuaishouFaster, cheaper Kling 3.0 at 1080p with native audio and improved lip-sync. No End frame or character elements.
Image models.
Nano Banana ProGoogleBest default for brand images and readable text.Nano Banana 2GoogleFast Google image generation.Nano Banana 2 LiteGoogleLow-latency Google image model for cost-efficient drafts.Seedream 5.0 ProByteDanceFlagship image generation with multilingual typography, dense layouts, and up to 10 edit references.Seedream 5.0 LiteByteDanceCost-efficient high-resolution image generation and editing with up to 10 references.Seedream 5.0 FlashByteDanceFast, low-cost Seedream 5.0 image generation and editing with up to 10 references.Meta Muse ImageMetaMeta Muse image generation and editing at one flat low price, with up to 10 edit references.GPT Image 2.5 FlareOpenAIGeneral-purpose OpenAI GPT Image 2.5: fast, natural lighting, strong typography; edits accept up to 16 references.GPT Image 2.5 SunburstOpenAIPrecision OpenAI GPT Image 2.5 tier: extra fidelity on intricate detail at longer generation time.Grok Imagine 2.0xAIxAI Grok Imagine 2.0 image model; edits accept up to 3 reference images.
Studio picks. You can too.
Studio sends each request to the model that handles it best, and you can always choose your own.