November 27, 2023

OctoAI launches it’s new Text Gen Solution with Llama 2, Code Llama, and Mistral models.

  • Run inference against multiple sizes of Llama2 Chat, Code Llama Instruct, and Mistral Instruct all via one unified API.

  • Reliably scale your app with OctoAI, which is already processing millions of inferences daily.


November 20, 2023

SDXL image generation has greatly been improved with the addition of SSD-1B, a distilled SDXL that is 50% faster, and LCM-LoRA, which enables image generation in less than 1 second.

  • SSD-1B is an OSS distilled SDXL that is 50% faster due to is smaller size. OctoAI applied our proprietary ML compiler to the model, and are able to generate SDXL images in 1.4 seconds using SSD-1B.

  • LCM-LoRA is a custom asset that enables high-quality image output requiring as few as 4 steps to generate images in less than 1 second.


November 16, 2023

Image Gen Solution fine-tuning for SD 1.5 and SDXL via API or web UI now available.

  • Customers can create their own unique fine-tunes in the web UI with at least 3-6 images, or import custom assets from popular sources to generate one of a kind images.

November 8, 2023

OctoAI Image Gen Solution launched.

  • The fastest and most customizable GenAI stack for production-grade image generation applications. The new solutions boasts SDXL images in less than 3 seconds, the ability to create or import fine-tuning assets, and reliable scale from a few to thousands of images.