2 min read

Chat with RTX, New Image Generation Model Stable Cascade, Gemini Ultra vs GPT-4 — Top AI News of the Week

Our latest AI Digest covers the biggest breaking AI news for the week. Anywhere Club community leader, Aliaksei Kartynnik, comments on key stories.

Anywhere Club community leader, Aliaksei Kartynnik

#1 — Local Chatbot from NVIDIA

NVIDIA has released a free genAI chatbot, Chat with RTX, which enables local deployment of an LLM on computers with NVIDIA 30/40-series graphics cards and at least 16GB of RAM. Using Chat with RTX you can interact with documents on your computer via a handy chatbot. "Interact with documents" means "conversing" with nearly any text documents or notes, preparing transcripts of YouTube videos, and even accomplishing some programming tasks. Basically, it offers the privacy and security of a personalized chatbot that resides on your own computer.

#2 — New Image Generation Model from Stability AI

Stability AI has introduced, in research preview, its new open-source text-to-image model. Stable Cascade is built on Würstchen architecture. It operates at the Stable Diffusion 2 level and surpasses Stable Diffusion XL in some aspects. Stable Cascade has a simple fine-tuning mechanism and is capable of generating images from prompts, suggesting generation variants, and generating images from references. Early reviews note that it is easy to train and customize.

#3 — Gemini Ultra vs GPT-4

There are already comparisons between Google's new Gemini Ultra and GPT-4 from OpenAI. How do they stack up? Neither is perfect, and both have their strengths. It looks like Gemini falls short in terms of response quality but surpasses GPT-4 in response speed. Gemini does not have rigid message count restrictions (unlike ChatGPT), and it demonstrates multimodal capabilities; it can generate images and audio using a single generalized model. You can try Gemini Ultra with a subscription to Gemini Advanced. And yes, the Gemini brand has completely replaced Bard. Some say this is Google's attempt to shake off Bard's tarnished reputation. We’ll see how that works out.

Bonus!

In these AI digests, a bonus tradition has taken root where we, as speakers, share something useful or amusing and the Anywhere Club blog editor turns a blind eye. My share for today is AI DIAL. This is EPAM's open-source LLM orchestrator that allows you to integrate various LLMs into your products through a uniform API. AI DIAL can work with multiple cloud providers via a single interface, and it offers a wealth of additional resources such as Helm charts, usage analytics, and authorization.