AI Video & Image Models
One page per model. Real pricing, real capabilities, no marketing fluff. Use this as the index for everything Slates supports plus the models worth knowing about even if we don't run them yet.
Seedance 2.0: ByteDance's Audio-First AI Video Model
→Seedance 2.0 is ByteDance's AI video model. It generates 4 to 15 second clips with synced audio baked into the same generation pass, at 480p, 720p, 1080p or native 4K. Raw provider pricing runs from $0.07 per second at 480p to $0.78 at 4K, far cheaper than CapCut Pro's subscription markup. Slates ships it on credits.
Kling AI: The Camera-Control King of AI Video
→Kling AI is Kuaishou's video generation model and the most camera-aware AI video tool on the market. V3 Standard runs at $0.084 per second, V3 Pro at $0.112, and V3 Omni at $0.084 with audio as a paid extra rather than a freebie. All three tiers do 15 second clips and 6-axis camera controls without any subscription.
Google Veo 3 / Veo 3.1: The Cheapest Native 4K AI Video Model
→Veo 3 (officially Veo 3.1) is Google's flagship AI video generation model, and the cheapest way to reach native 4K per second. Raw API pricing starts at $0.10 per second for the Fast tier without audio and runs up to $0.60 per second for Standard 4K with audio. Slates ships it on credits next to Kling and Seedance.
Nano Banana 2: Google's Fast, Cheap Image Generation Model
→Nano Banana 2 is Google's image generation model. The technical name is Gemini 3.1 Flash Image. It replaced the previous Nano Banana Pro on February 26, 2026, and runs about half the price, at $0.08 per image at 1K resolution. It's built on the Gemini Flash architecture, so generation latency is sub-500ms.
Seedream 5 Lite: ByteDance's Uncensored Image Model
→Seedream 5 Lite is ByteDance's image generation and editing model. It's the uncensored fast-and-cheap option in most workflows, with strong face consistency and the ability to reference up to 10 images per edit. It ships in Slates on credit-based pricing and is the default fallback when Nano Banana 2 rejects a prompt.
Sora 2: What It Was, What It Cost, and What To Use Now That OpenAI Is Retiring It
→Sora 2 is OpenAI's text-to-video model, released September 30, 2025, and OpenAI is retiring it: the Sora app closed on April 26, 2026, and the Sora 2 API is scheduled for removal on September 24, 2026. For the same audio-first use case with a future, Slates runs Seedance 2.0, which generates video and audio in one pass and goes all the way to 4K.