Community provider for the AI SDK (ai 7). One Cut.Pro key gives generateImage, generateSpeech and experimental_generateVideo access to Seedance 2.5, Wan 3, Kling O3, Veo 3.1, Gemini Omni Flash, Nano Banana Pro, Seedream 5, Qwen Image, GPT Image, Grok Imagine, MiniMax and Gemini TTS voices. Priced per second of video, per image or per 1,000 characters, in reais and dollars: see the live price list. Pay with Pix or card; companies get an invoice.
npm install ai @cutpro/ai-sdk-providerCreate a key at cut.pro/studio/me/api-keys (every account starts with free credits) and set CUTPRO_API_KEY.
import { generateImage } from "ai";
import { cutpro } from "@cutpro/ai-sdk-provider";
const { image } = await generateImage({
model: cutpro.image("google/nano-banana-pro"),
prompt: "a ceramic coffee mug on a marble counter, morning light, product photo",
aspectRatio: "4:5",
});With reference images (links or bytes):
const { image } = await generateImage({
model: cutpro.image("google/nano-banana-pro"),
prompt: {
text: "the person from the first image wearing the jacket from the second",
images: ["https://your-cdn.com/person.jpg", "https://your-cdn.com/jacket.jpg"],
},
});import { experimental_generateVideo as generateVideo } from "ai";
import { cutpro } from "@cutpro/ai-sdk-provider";
const { video } = await generateVideo({
model: cutpro.video("bytedance/seedance-2.5"),
prompt: "a golden retriever surfing a wave at sunset, cinematic",
aspectRatio: "16:9",
resolution: "1280x720",
duration: 5,
});Image to video with Wan 3: prompt: { image: "https://your-cdn.com/product.jpg", text: "the camera slowly orbits the product" } and cutpro.video("alibaba/wan-3"). First and last frame: frameImages. The SDK polls until the video is ready (default timeout 10 minutes, set poll: { timeoutMs } for longer ones).
Video generation is experimental in the AI SDK itself (experimental_generateVideo, Experimental_VideoModelV4), so its interface may change between AI SDK releases.
import { generateSpeech } from "ai";
import { cutpro } from "@cutpro/ai-sdk-provider";
const { audio } = await generateSpeech({
model: cutpro.speech("google/gemini-3.1-flash-tts"),
text: "Welcome to Cut.Pro. Your video is ready.",
voice: "Kore",
});
// audio.uint8Array is an MP3Voice ids per model are in GET https://api.cut.pro/api/v1/models?id=google/gemini-3.1-flash-tts (voices[].id), or with @cutpro/sdk: cutpro.models.list({ id }).
| AI SDK | Cut.Pro |
|---|---|
prompt |
prompt |
aspectRatio |
aspect_ratio (21:9, 16:9, 3:2, 4:3, 5:4, 1:1, 4:5, 3:4, 2:3, 9:16) |
resolution: "1280x720" (video) |
resolution: "720p" |
duration |
seconds |
generateAudio |
audio |
seed |
seed |
image prompt.images |
mode: "references" with image_urls (bytes are uploaded first) |
video prompt.image / frameImages / inputReferences |
mode: "image" / "frames" / "references" |
voice |
voice |
providerOptions.cutpro |
any other field of POST /generations: resolution, mode, preset, options, audio_urls, video_id, webhook_url, credits |
Not supported (a warning is returned): image size and mask, video fps, speech speed, instructions, language and formats other than MP3. Each call generates one image or video: n: 4 makes four generations.
Every call quotes the request and agrees to the quoted total before generating. To cap spending, quote first with @cutpro/sdk and pass providerOptions: { cutpro: { credits } }: if the price changed, nothing is charged and the call fails with PRICE_CHANGED. Errors are CutProError with code, status and extra (see the error catalog). The provider downloads the full quality file, after which the credits of that generation can no longer be asked back.
providerMetadata.cutpro carries the generation id, the credits charged and the seed.
import { createCutPro } from "@cutpro/ai-sdk-provider";
const cutpro = createCutPro({
apiKey: process.env.CUTPRO_API_KEY,
workspaceId: "...", // only for a key that reaches more than one workspace
baseURL: "https://api.cut.pro/api/v1",
fetch: customFetch,
});Any other image, video or voice model id from the catalog works too.
Provider da Cut.Pro para o Vercel AI SDK: generateImage, generateSpeech e experimental_generateVideo com Seedance 2.5, Wan 3, Kling O3, Veo 3.1, Nano Banana Pro, Seedream 5, GPT Image e vozes do Gemini TTS, com uma chave só. Preço por segundo de vídeo, por imagem ou por 1.000 caracteres, em reais: tabela de preços ao vivo. Pague com Pix ou cartão, com nota fiscal para empresas. Toda conta começa com créditos grátis.
import { generateSpeech } from "ai";
import { cutpro } from "@cutpro/ai-sdk-provider";
const { audio } = await generateSpeech({ model: cutpro.speech("google/gemini-3.1-flash-tts"), text: "Olá! Seu vídeo está pronto.", voice: "Kore" });Crie a chave em cut.pro/studio/me/api-keys. Páginas dos modelos: Seedance 2.5, Wan 3, Nano Banana Pro, Gemini 3.1 Flash TTS.
MIT