{"ok":true,"trend":{"id":878846,"platform":"mastodon","region":"global","key":"🐦 # alephalpha released # kolibri , an open-weight # llm for german and english under the apache 2.0 license. a # mixtur","title":"🐦 # AlephAlpha released # Kolibri , an open-weight # LLM for German and English under the Apache 2.0 license. A # Mixtur","url":"https://social.vivaldi.net/@michabbb/117377673574532261","first_seen":"2026-10-03T16:20:17.889277Z","last_seen":"2026-10-03T16:20:17.889277Z","last_rank":6,"peak_rank":6,"last_volume":1,"peak_volume":1,"seen_count":1,"score":0.98171875,"category_hint":"software","section":"technology","category":"software","summary":"German AI company Aleph Alpha has released Kolibri, an open-weight large language model for German and English under the Apache 2.0 license. The Mixture-of-Experts model has 78.1 billion parameters, with only 3.46 billion active per token, and was trained from scratch on roughly 24 trillion tokens, more than a fifth of them in German, using 768 NVIDIA B200 chips. Observers highlight it as a notable European, permissively licensed alternative to US models.","why":"A major European AI lab has released a powerful openly licensed model with strong German-language capability, which is rare in the field.","tone":"positive","entities":["Aleph Alpha","Kolibri","NVIDIA","Germany"],"summarized_at":"2026-10-03T16:21:35.493542Z","meta":{"tag":"opensource","via":"scan","kind":"status","lang":"en","instance":"mastodon.social","tag_uses":691},"nw":null,"promo":null,"kind":null,"importance":null,"hidden":false,"hide_reason":null,"judged_at":null,"title_en":"Aleph Alpha releases Kolibri open-weight bilingual AI model","section_name":"Technology","category_name":"Software","timeline":[{"captured_at":"2026-10-03T16:20:17.889277Z","rank":6,"volume":1}],"posts":[{"platform":"mastodon","url":"https://social.vivaldi.net/@michabbb/117377673574532261","author":"michabbb@vivaldi.net","title":null,"snippet":"🐦 # AlephAlpha released # Kolibri , an open-weight # LLM for German and English under the Apache 2.0 license. A # MixtureOfExperts with 78.1B parameters, of which only 3.46B are active per token. Trained from scratch on ~24T tokens (over a fifth German) using 768 # NVIDIA B200…","posted_at":"2026-10-03T15:26:01Z","likes":1}],"elsewhere":[],"window":"7d"}}