We would like to provide an assessment here. However, it is extremely important to us to point out the following. The information provided on this website does not constitute legal advice and is not intended to address legal questions or issues that may arise in individual cases. The information on this website is of a general nature and is provided for informational purposes only. If you require legal advice for your individual situation, you should seek the advice of a qualified lawyer. In addition, this technology is highly volatile, so the following must be viewed in context. We updated this information in April 2025.

Brief description and distinction between ChatGPT and Mistral AI

The rapid development of large language models (LLMs) enables companies to use a wide variety of models as tools. As of April 2025, the leading developers in this field include OpenAI, the creator of Chat GPT, and Mistral AI, a prominent European competitor.

LLMs differ significantly in their technological approaches, business models and strategic orientations. In the following, we will limit ourselves to a consideration of ChatGPT and Mistral from the daily growing variety. And we will refer to the use case in contact centres via voice/chat bots in Europe. We will attempt to provide some clarity regarding the underlying technologies, performance characteristics, licensing structures and possible applications.

API (Application Interface)

We are pursuing a structural approach to connect the jtel portal (bot application) with a public LLM model.
In very basic terms, one could draw an analogy between IT (information technology) and AI (artificial intelligence). In this possible comparison, the IT operating system (Linux, Microsoft) would correspond to the AI LLM (Mistral, ChatGPT), on which we build valuable applications (bots). The jtel bots therefore interact with these third-party AI language models via interfaces (APIs). 

ChatGPT

ChatGPT is primarily a generative chatbot application developed by OpenAI in 2022 (headquartered in San Francisco, USA – ‘capped profit’ structure with a type of non-profit status) with a focus on a user-friendly interface, which has led to widespread adoption among end users. Of course, the central service also provides the necessary API interfaces to communicate with the jtel platform. The commercial GPT (Generative Pre-trained Transformer) model series currently includes GPT-4o (o1, o3-mini) and GPT-4.5. OpenAI's models are trained on huge data sets that include publicly available Internet information, data licensed from third parties, and data generated by users or human trainers. Microsoft is a major investor in OpenAI and provides extensive cloud computing resources via Azure. This explains the location in the USA, where data processing takes place under American law. It is worth noting that OpenAI is under scrutiny for possible copyright infringements and AI security practices.

Mistral AI

Mistral AI SAS, on the other hand, focuses on the API platform ‘La Plateforme’ and architectural efficiency. It was founded in 2023 (headquartered in Paris, France) with the aim of offering an alternative through an open-source model. The Mistral AI model series includes a variety of ‘general-purpose,’ ‘specialist,’ and ‘research models’ as a service or open-source variant for operation on your own hardware (self-hosting). Optimised commercial models are currently available via the API platform as Mistral Large, Mistral Small, and other specialised models.
Mistral AI has secured significant funding from various investors, including Lightspeed Venture Partners, Andreessen Horowitz, Microsoft, Nvidia and BNP Paribas. The company has also entered into a partnership with Microsoft to make its models available on the Azure cloud platform. The company emphasises its European roots and its commitment to compliance with the General Data Protection Regulation (GDPR) as an important distinguishing feature.

Personal assessment:

While OpenAI's largest models often lead the general benchmarks, Mistral AI is a strong competitor, especially Mistral Large. Mistral's efficiency-focused architectures may offer an advantage in speed with comparable capabilities. For European companies, the more efficient architecture and specific focus on key European languages may be a decisive factor. Benchmarks seem to show that Mistral Large models such as Llama 2 70B excel in French, German, Spanish and Italian. This suggests that Mistral could offer superior quality, accuracy and cultural nuance for applications targeting these specific language markets.

In addition, Mistral allows self-hosting of open models (which run under Apache 2.0 license, such as Mistral 7B and Mixtral). These can be downloaded and deployed on a company's own infrastructure (on-premise servers or the jtel SaaS Private Cloud). This use case thus reliably prevents the transfer of any personal data collected during conversations outside the European Economic Area (EEA). 

Mistral's European base and its open-source philosophy provide a foundation for verifiable compliance with the GDPR. The self-hosting option for open models also offers maximum data control and sovereignty, directly addressing the most important GDPR concerns regarding data storage location and third-party access. An EU-based infrastructure further strengthens jtel's data residency capabilities for its LLM API services. 

Depending on what data is exchanged with the LLMs via the jtel Platform API, whether customer-specific data is included in the conversation or only individual key information is transferred via corresponding ‘prompts,’ it may be more or less relevant to use a European Mistral service or even a local Mistral instance.

These are important building blocks for facilitating GDPR compliance. However, any company using our APIs to LLMs must still exercise due diligence by reviewing local GDPR data processing practices on a case-by-case basis.