--- title: "Seedance 2.5 and 2.0 Are Now Live on Yotta AI Gateway" slug: seedance-2-5-and-2-0-yotta-ai-gateway-video-api description: "ByteDance's Seedance 2.5 and 2.0 are now on the Yotta AI Gateway. Text to video and image to video, up to 30s clips, priced per second." author: "Yotta Labs" date: 2026-08-31 categories: ["News"] canonical: https://www.yottalabs.ai/post/seedance-2-5-and-2-0-yotta-ai-gateway-video-api --- # Seedance 2.5 and 2.0 Are Now Live on Yotta AI Gateway ![](https://cdn.sanity.io/images/wy75wyma/production/ed72a6707e104f024f65aa088bf7481172639f5f-1200x627.png) *ByteDance's current Seedance family joins the Gateway's video lineup. Here's what shipped, what it costs, and how to send your first request.* The Yotta AI Gateway now serves ByteDance's Seedance 2.5 and Seedance 2.0, each available in text-to-video and image-to-video form. They join the Gateway's existing video catalog, which already includes Wan 3.0, Wan2.7, [HappyHorse-1.0](https://www.yottalabs.ai/post/what-is-happy-horse-1-0-the-new-ai-video-model-explained-2026), and Seedance 1.5 Pro, and they run through the same account, API key, and billing as everything else on the Gateway. The addition matters because Seedance 2.x is where ByteDance's video work currently lives: the models behind this year's wave of interest in AI video, now callable next to the rest of the catalog. The same key that routes your chat completions can turn a product shot into a 15 second clip or a text prompt into a 30 second scene with audio. ## **What's in the catalog** Four models went live: Seedance 2.5 is ByteDance's newest video model, released July 31, 2026. It generates up to 30 seconds in a single request, supports multi-round extension for longer content, and handles timestamp-precise editing instructions in the prompt. ByteDance's own positioning is long-form storytelling: one generation that holds a narrative together instead of stitched 5 second fragments. Seedance 2.0, released in February 2026, remains the workhorse. It accepts multimodal input, holds character consistency across shots, and is the only one of the two that outputs 4K on the Gateway. It's also cheaper per second at every resolution. For the full head-to-head, see our [Seedance 2.5 vs 2.0 comparison](https://www.yottalabs.ai/post/seedance-2-5-vs-seedance-2-0-differences-which-to-use-2026). ## **Pricing** Video on the Gateway is billed per second of generated output, priced by resolution. Audio generation doesn't change the price. Concretely: a 10 second 720p draft on Seedance 2.5 costs $2.40. The full 30 second 1080p version costs $12.60. An 8 second 720p clip on Seedance 2.0 costs $1.28. The pattern that keeps bills sane is iterating at 480p or 720p and rendering the final cut at full resolution, which the per-second pricing makes easy to reason about. Current rates are always on the [pricing page](https://www.yottalabs.ai/pricing). ## **How it works** Video generation is asynchronous. You submit a job, get a request ID back, and poll for the result: ```bash curl -X POST "$BASE_URL/text-to-video/generations" \ -H "X-API-KEY: $YOTTA_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "model": "seedance-2.5-t2v", "parameters": { "prompt": "A paper boat drifting down a rain-swollen street gutter, cinematic", "duration": 8, "resolution": "720p", "aspect_ratio": "16:9", "generate_audio": true } }' ``` Poll GET /text-to-video/generations/{request_id} until the status is completed and you get back an output URL. Image to video works the same way at the /image-to-video/generations endpoint, with a first_frame image URL and an optional last_frame to control where the clip starts and ends. Prompts can run long. Seedance rewards detailed direction: shot lists, timing cues, audio design, and camera moves, up to roughly 1,000 words. Duration also has a smart mode: pass -1 and the model picks a length that fits the prompt. The step-by-step version, including a Python client that submits and polls for you, is in the [Seedance API guide](https://www.yottalabs.ai/post/how-to-use-seedance-api-generate-video-2026). ## **Why this matters for Gateway users** The Gateway already gives you one API across GLM, DeepSeek, Qwen, Kimi, and the rest of the text catalog, plus the existing video and image models. Seedance 2.5 and 2.0 add the current ByteDance flagship to that lineup, so the marketing clip, the product demo, and the LLM pipeline that writes the script for both can live behind one key with one bill, instead of a separate video vendor with separate auth and separate invoicing. And with Wan 3.0 at $0.035/s and Seedance 1.5 Pro at $0.012/s sitting in the same catalog, you can route drafts to a cheaper video model and finals to Seedance 2.x the same way you already route between text models. And because billing is unified, video spend shows up in the same usage dashboard as your token spend, so cost tracking per project or per feature keeps working. ## **Get started** Sign in to the [console](https://console.yottalabs.ai), grab an API key, and the Seedance models are in the model library alongside everything else. If you're new to the Gateway, the [AI Gateway page](https://www.yottalabs.ai/ai-gateway) covers how routing, keys, and billing work. ## **Frequently asked questions** **Which Seedance model should I start with?** Seedance 2.0 for drafts and 4K finals, Seedance 2.5 for anything longer than 15 seconds or with complex storytelling. The [full comparison](https://www.yottalabs.ai/post/seedance-2-5-vs-seedance-2-0-differences-which-to-use-2026) breaks down when each wins. **How is video billed?** Per second of generated video, priced by resolution. A 10 second 720p Seedance 2.5 clip is 10 x $0.24 = $2.40. Turning audio on or off doesn't change the rate. **Can I generate video from an image?** Yes. Both models have image-to-video variants (seedance-2.5-i2v, seedance-2.0-i2v) that take a required first frame and an optional last frame. **Is the API synchronous?** No. Video jobs are async: submit, receive a request ID, poll for status. Longer, higher-resolution jobs take longer to render. **Do videos have watermarks?** Watermarking is a request parameter and defaults to off on the Gateway. **Does my existing Gateway API key work?** Yes. Video endpoints use the same key and the same billing account as the rest of the Gateway catalog. Setup details are in the [API guide](https://www.yottalabs.ai/post/how-to-use-seedance-api-generate-video-2026).