NinjaTech releases SuperGPT: Supercharging AI Assistant with Llama3.1 405B

NinjaTech's latest update, SuperGPT, integrates Llama 3.1 405B, FLUX.1 models, to supercharge its AI Assistant capabilities

It’s been less than two months since Meta launched Llama 3.1 405B, its super-sized LLM that outperforms competing models like GPT4-o and Claude 3.5 Sonnet on a variety of benchmarks. Already, fast-moving startups in the Gen AI space are putting 405B into production.

One that caught my eye this week is NinjaTech’s “SuperGPT”. Ninjatech builds user-friendly tools for designing and deploying AI agents for tasks like image generation and editing, coding, professional and creative writing, deep research, meeting scheduling, file-upload comprehension and analysis, and multi-modal image analysis. With SuperGPT, NinjaTech has announced a slew of updates, built around the integration of Llama 3.1 405B and new image models into its AI assistant.

NinjaTech AI Assistant, Currently in Beta

At the heart of the update is NinjaTech's state-of-the-art agentic model, Ninja-LLM 3.0. Fine-tuned using Meta's Llama 3.1 405B, the company’s analysis of 750+ real world queries showed this high-performance, quantized model is competitive with state-of-the-art models like Claude Sonnet 3.5. In effect, with Llama 3.1 405B, Ninjatech’s AI Assistant just got a whole lot smarter. Ninjatech’s vision for SuperGPT is that it will “become a digital coworker that you can assign complicated tasks to asynchronously and solve them conversationally, saving you time and money.”

New Image Models

The update also includes unlimited image generation and editing through two new first-party models: Ninja Pix and Ninja Pix Pro. Based on PixArt-Σ and Flux.1, respectively, these models offer unparalleled capabilities for text-to-image generation. Ninjatech provides unlimited use for these new models, although it does not appear to support image fine-tuning.  FLUX.1 from Black Forest Labs is the latest state-of-the-art diffusion model for image creation. You can read my full write-up about it here.

Image Recognition

Another interesting new feature introduced in the release is MyNinja multimodal capabilities, which offer the ability to comprehend photos. By giving its AI assistant a pair of eyes, users can now interact with MyNinja by taking a picture. Use cases for image recognition include tasks such as translating text, setting up calendar events, finding comparable products, or reviewing code based on the contents of the photo.

New Desktop Apps

To further unlock the power of MyNinja, NinjaTech has also announced the launch of native desktop apps for MacOS, with Windows, Linux, and mobile apps coming soon. This means that users can access MyNinja from anywhere, at any time, to boost their productivity.

All these features are available for just $10/month. It’s a tempting offer. As someone who’s been learning to build AI agents in Langchain recently, it is no easy feat and remains mostly the domain of sophisticated developers. For the cost of a couple lattes, you can skip the coding and unlock the power of NinjaTech's SuperGPT AI Assistant to start building your own AI agents. With over 1M monthly users, the company is clearly on to something here.

In an era where AI is increasingly becoming a part of our daily habits, NinjaTech's latest update is a game-changer. By democratizing access to the world's best LLM and image models and creating tools to make them useful in our daily lives, the company may be on a fast-track to creating the next killer productivity app for Gen AI.

You can sign up for a free 7-day trial of MyNinja here and start experiencing the future of AI-assisted productivity for yourself.