Přístupnostní navigace
E-application
Search Search Close
Guide to computing, data, development and experimental capacities for working with artificial intelligence at BUT. The site summarizes what tools and infrastructure students, staff, research teams and collaborative partners can use for AI chat, programming, data analysis, training and inference of models, deploying applications, working with research data and testing industrial AI solutions.
On this page
The AI@BUT portal is listed as one of the available services GPT Chat Open WebUI within e-INFRA CZ. The service provides on‑premise access to language models for conversations, programming, working with documents and other tasks. The e-INFRA/CERIT‑SC documentation states that Open WebUI requires a valid MetaCentra account and that the same models can also be used via the documented REST API.
https://llm.ai.e-infra.cz/v1/
Always use the current service documentation for model comparisons, available model names, and the exact login procedure. The model list is subject to change based on hardware availability, testing and operational priorities.
It is suitable for teaching, data analysis, research scripts and prototyping AI workflow JupyterHub in the e-INFRA CZ environment. The documentation lists support for Jupyter Notebooks, JupyterLab, RStudio, and ready-made Docker images, including images for TensorFlow, PyTorch, Matlab, R, and AI assistance.
At BUT, there is also a teaching license for programming GitHub Copilot for students and teachers. Before using Copilot or another assistant in a project, check the license terms, non-public code rules, partner agreements, and repository rules.
For demanding calculations, e-INFRA CZ offers the most powerful and modern supercomputer systems in the Czech Republic, grid and cloud infrastructure for demanding calculations and extensive data sets. The e-INFRA public page lists IT4Innovations, MetaCentrum Grid, MetaCentrum Cloud, Kubernetes, AI tools and SensitiveCloud among the categories of Supercomputers.
MetaCenter provides computing and storage capacities, specialized hardware, software, network infrastructure and user interfaces such as OnDemand, Galaxy or Rancher. The grid approach is suitable for batch jobs from hours to weeks; the cloud approach is suitable if the project needs its own virtual machines or a specific environment.
IT4Innovations is a national supercomputing center. A supercomputer Caroline is designed for scientific and industrial problems including big data analytics and artificial intelligence, according to public data from IT4Innovations. Karolina has a theoretical performance of 15.7 PFlop/s, an accelerated part with 72 servers of 8 GPU accelerators and up to 360 PFlop/s for artificial intelligence calculations; fast data storage provides more than 1.4 PB with speeds up to 1 TB/s.
IT4Innovations distinguishes several categories of access. Public grant competition, institutional and community access, and rapid and testing access are especially important for research teams. According to IT4Innovations, the public grant competition is announced three times a year, usually in February, June and October. The fast and test approach is intended for smaller short-term tasks, concept testing, debugging and preparation for a larger request.
AI projects usually need reliable work with data: secure transfer, team sharing, temporary work storage, long-term archiving, repository for published data and data workflow documentation. The e-INFRA CZ documentation guide lists, among others, Object Storage, the National Repository Platform, ownCloud and FileSender.
Also use methodological support for research data Open Science at BUT. For AI datasets, it is appropriate to document data origin, license, consents, anonymization, metadata description, dataset version, pre-processing, training/validation/testing and restrictions on further use.
The official statement of BUT states that before data is provided to generative AI tools, the conditions for the protection of input data must be verified and that non-public information from activities where BUT is contractually obligated to protect the rights and know-how of third parties must not be provided to generative AI tools.
At the same time, the CERIT‑SC documentation distinguishes between the common use of Open WebUI and a more sensitive workflow. It states that standard AI services run in the infrastructure of e-INFRA CZ and that questions and answers are not forwarded to external providers by default; at the same time, however, it explicitly points out that WebUI is not certified as an environment for protected, classified or personal data, and for sensitive data it is necessary to choose a more appropriate mode, such as direct API, specialized deployment or SensitiveCloud after consultation.
BUT has top research infrastructure, which is also important for applied AI: it enables data acquisition, model verification in specific domains, robotics, measurement, imaging, industrial testing, materials research, energy or smart infrastructure. Public page Research infrastructures - BUT states that BUT is the host institution of CzechNanoLab and a partner of Czech‑BioImaging and that other modern infrastructure is concentrated in centers of top research infrastructure.
CEITEC BUT is part of the European network AI‑MATTERS, which connects Testing and Experimentation Facilities for industrial AI. The Czech team includes CIIRC CTU, CEITEC BUT and VŠB-TUO. The services are mainly intended for testing AI software and hardware in a real or realistic industrial environment.
According to public information from BUT and CEITEC, RICAIP Testbed Brno can offer experiments on infrastructure simulating an industrial environment. The mentioned equipment includes CNC machines, laser cutters or welding centers, 3D printing of metals and plastics, robotic arms, communication technology including a private 5G network and a powerful computing cluster or supercomputer for demanding tasks.
BUT is part of the consortium Czech AI Factory. According to the BUT report from October 2025, the new KarolAIna supercomputer is to be installed in the IT4Innovations national supercomputer center in Ostrava within the next three years. The resulting infrastructure is intended to serve industry, public administration, research teams, and startups, and is to be complemented by the consortium's expert services, such as consulting, models, and algorithms.
Take this part as strategically prepared infrastructure, not as a commonly available service at the time this page was created. For current access options, follow the websites of BUT, AI@BUT, IT4Innovations, EuroHPC JU and Czech AI Factory.
Responsible person: doc. Ing. Petr Koňas, Ph.D.
Responsibility: doc. Ing. Petr Koňas, Ph.D.