ChatMDR 3 and introducing ChatIVDR

ChatIVDR

Often considered the poor cousin of the Medical Device Regulation (MDR), the In-Vitro Diagnostic Device Regulation or IVDR did not get as much attention.

Until now! We are proud to introduce ChatIVDR, an AI chatbot based on OpenAI ChatGPT technology that incorporates the well-established technology of ChatMDR. ChatIVDR incorporates the IVDR legislation as well as relevant MDCG guidance documents and standards.

ChatIVDR is the only AI tool in the market today that serves in-vitro diagnostic device manufacturers.

You can use ChatIVDR with the same account as ChatMDR. You don’t need an extra subscription!

Read all about ChatIVDR here.

Smarter AI model GPT4-o for all customers, including the Free Plan

The best AI model, based on ChatGPT 4 which is the best AI large language model in the world, was only available to customers on the ChatMDR Plus and ChatMDR Unlimited plans.

Starting with ChatMDR 3, the most powerful GPT4 model is also available to the customers on the Free Plan! The only restriction of the Free Plan is that you can only ask 20 questions per month to ChatMDR (or ChatIVDR) and that the maximum chat length is three questions.

Actually, there is a second announcement here. ChatMDR 2 was based on GPT 4 Turbo, the most advanced model at the time. Our partner OpenAI released their latest model called GPT4-o on May 13 and it took the top spot in the AI leaderboard.

The “o” stands for “omni”, which means that the model can now also directly work with images, video and sound without first having to convert it to text. For ChatMDR and ChatIVDR, it is “only” the best AI model available right now, but wait and see what we will be able to do with the other capabilities in future!

Speedy llamas

You just read about the Smarter model in ChatMDR and ChatIVDR version 3, which runs on OpenAI’s GPT4-o. This model is already 50% faster than ChatMDR version 2.

For the Faster model, we went a different route. There, we use LLama3-70B running on Groq.

Huh?

Yeah, that probably requires some explanation. LLama3 is a new AI language model by Meta, the company behind Facebook. Released in April 2024, it now powers the faster of the two available models. And it is…really fast! Go see for yourself.

Privacy

You read “Facebook” there, but do not worry. No ChatMDR/ChatIVDR model stores your data or uses it for future model training! Your secrets are safe with us. Even the LLama3 model does not actually run at Meta/Facebook, but at our supplier Groq, who have the fastest AI serving technology on the market. Neither Groq, not ChatMDR stores your data. Here’s an excerpt from the Groq terms of use:

If you supply any Training Data for the purpose of prompting, fine-tuning or customizing the Generative AI Services to your specific needs or use-case, Groq will not use the Training Data other than to perform the Generative AI Services for you. Groq does not permanently retain Prompts, Output or your Training Data on its servers.