{"id":"minicpm-o-4-5","nom":"MiniCPM-o 4.5","editeur":"OpenBMB","pays":"Chine","page":"https://quelleia.com/voix/modeles/minicpm-o-4-5/","domaine":"voix-a-voix","annonce_le":"2026-05-17","en_attente":true,"mesure":null,"classe":false,"note":"pas encore mesuré par une source publique ; aucun rang, aucun chiffre de classement","classements":[],"poids_ouverts":true,"licence":"Apache 2.0","liens":{"poids":"https://huggingface.co/openbmb/MiniCPM-o-4_5","page":"https://huggingface.co/openbmb/MiniCPM-o-4_5","essayer":"https://minicpmo45.modelbest.cn/"},"presentation":"Modèle omnimodal à poids ouverts d'OpenBMB qui écoute, regarde et parle en même temps, en duplex intégral, et peut prendre la parole de lui-même selon ce qu'il perçoit. Sa conversation vocale fonctionne en anglais et en chinois, avec des voix réglables et le clonage d'une voix à partir d'un court extrait.","taille":"9 milliards de paramètres","langues":{"annoncees":{"annoncees":"anglais et chinois","nombre":2,"francais":false,"liste":["en","zh"]}},"annonce_de_l_editeur":{"editeur":"OpenBMB","page":"https://huggingface.co/openbmb/MiniCPM-o-4_5","date":"2026-10-23","verifie":"2026-10-23","hors_classement":true,"chiffres":[{"nom":"Délai avant le premier jeton","precision":"en bf16 comme en int4 ; premier jeton, pas forcément premier son ; matériel non précisé","valeur":0.6,"unite":"s","provenance":"calculé par OpenBMB"}]},"preuves":[{"appuie":["presentation","taille"],"lu":"https://raw.githubusercontent.com/OpenBMB/MiniCPM-o/main/README.md (« MiniCPM-o 4.5 is the latest model in the MiniCPM-o series, a **9B end-to-end omnimodal model**. It can continuously receive video and audio while generating speech and text, enabling real-time interactions that see, hear, and speak simultaneously. »), 23/10/2026"},{"appuie":["presentation"],"lu":"https://raw.githubusercontent.com/OpenBMB/MiniCPM-o/main/README.md (« Beyond responding to requests, the model can also interact proactively, for example by initiating reminders or comments based on its continuous understanding of the scene. »), 23/10/2026"},{"appuie":["presentation","langues"],"lu":"https://huggingface.co/openbmb/MiniCPM-o-4_5/raw/main/README.md (« MiniCPM-o 4.5 supports **bilingual real-time speech conversation with configurable voices** in English and Chinese. »), 23/10/2026"},{"appuie":["presentation"],"lu":"https://huggingface.co/openbmb/MiniCPM-o-4_5/raw/main/README.md (« The model also allows for fun features such as **voice cloning and role play via a simple reference audio clip** »), 23/10/2026"},{"appuie":["taille"],"lu":"https://huggingface.co/openbmb/MiniCPM-o-4_5/raw/main/README.md (« The model is built in an end-to-end fashion based on SigLip2, Whisper-medium, CosyVoice2, and Qwen3-8B with a total of 9B parameters. »), 23/10/2026"}]}