Inference Endpoints

Salad Inference Endpoints (SIE) is a powerful service provided by Salad Technologies that allows enterprises to run inference tasks using a variety of AI models. Designed specifically for enterprise-grade API products, SIE supports a range of applications, such as transcription and image generation.

SIE includes a user-friendly playground for testing and visualizing model outputs, automated testing and status reporting for product owners, and SDKs for seamless integration into customer applications.

Transcription

Salad Transcription is a powerful API that converts audio and video files into accurate transcripts, captions, or subtitles. It supports common formats like MP4, MOV, WAV, and MP3, and offers transcription in 97 languages, including English, Spanish, and Arabic. With features like speaker identification, accent differentiation, and sentence-level timestamps, it’s ideal for precise captioning or subtitling. Users can also generate SRT files compatible with popular video editors. The API integrates easily into existing platforms, with flexible options for translation, timestamps, and more.

To learn more visit our Transcription service documentation page.

Dreambooth Training

Salad's Dreambooth training service offers a cost-effective solution for fine-tuning Stable Diffusion models using reference images. Powered by hundreds of RTX 4090 GPUs on Salad’s distributed infrastructure, it provides up to 40% savings compared to other options. The service is ideal for companies needing large-scale AI training for tasks like AI avatars, product photos, and custom art styles. Using a simple orchestration API and GPU worker nodes, it ensures efficient handling of node interruptions and concurrent jobs, while leveraging serverless technologies to minimize operational overhead.

To learn more visit our Dreambooth documentation page.