{"id":167258,"date":"2026-04-03T05:11:55","date_gmt":"2026-04-03T03:11:56","guid":{"rendered":"https:\/\/prohoster.info\/blog\/novosti-interneta\/google-vypustil-otkrytuyu-ai-model-gemma-4-postroennuyu-na-tehnologiyah-gemini-3"},"modified":"2026-04-03T05:11:55","modified_gmt":"2026-04-03T03:11:56","slug":"google-vypustil-otkrytuyu-ai-model-gemma-4-postroennuyu-na-tehnologiyah-gemini-3","status":"publish","type":"post","link":"https:\/\/prohoster.info\/it\/blog\/novosti-interneta\/google-vypustil-otkrytuyu-ai-model-gemma-4-postroennuyu-na-tehnologiyah-gemini-3","title":{"rendered":"Google ha rilasciato il modello AI aperto Gemma 4, costruito sulle tecnologie di Gemini 3","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"

Google ha pubblicato una nuova famiglia di modelli linguistici di grandi dimensioni Gemma 4, basati sulla tecnologia del modello Gemini 3. Gemma 4 \u00e8 distribuito con licenza Apache in versioni da 2,3, 4,5, 25,2 e 30,7 miliardi di parametri (E2B, E4B, 31B e 26B A4B). Le varianti E2B ed E4B sono adatte per l'uso su dispositivi mobili, sistemi Internet of Things (IoT) e schede tipo Raspberry Pi, mentre le altre varianti sono idonee per workstation e sistemi con GPU consumer. La dimensione del contesto considerato dal modello \u00e8 di 128.000 token per i modelli E2B e E4B, e di 256.000 token per i modelli 31B e 26B A4B. <\/p>\n

Modelli multilingue e multimodali: supportano 35 lingue direttamente dalla scatola (sono stati utilizzati pi\u00f9 di 140 lingue per l'addestramento), e possono elaborare testo e immagini in ingresso (i modelli E2B ed E4B supportano inoltre l'elaborazione del suono). Il modello 26B A4B si basa su architettura MoE (Mixture-of-Experts), nella quale il modello \u00e8 suddiviso in una serie di reti esperte (in fase di generazione della risposta possono essere utilizzati solo 3,8 miliardi di parametri, ma la velocit\u00e0 \u00e8 notevolmente superiore rispetto ai modelli grandi classici), mentre gli altri modelli utilizzano un'architettura monolitica classica. <\/p>\n

I modelli supportano ragionamenti e modalit\u00e0 di pensiero personalizzabili, consentono un ruolo sistemico (System Role) per elaborare istruzioni (regole, limitazioni) separatamente dai dati. I modelli possono essere utilizzati per scrivere codice, riconoscere oggetti nelle immagini, analizzare video fotogramma per fotogramma, elaborare documenti e PDF, riconoscere testo stampato e manoscritto (OCR), riconoscere la voce e tradurre tra lingue. \u00c8 possibile utilizzarli come agenti autonomi, interagendo con vari strumenti e API. <\/p>\n

Nei principali test, i modelli della serie Gemma 4 hanno superato significativamente il modello Gemma 3 con 27 miliardi di parametri. L'uso di Gemma 4 \u00e8 supportato da strumenti e librerie come LiteRT-LM, vLLM, llama.cpp, MLX, Ollama, NVIDIA NIM e NeMo, LM Studio, Unsloth, SGLang, Cactus, Basetan, MaxText, Tunix e Keras.

\"Google<\/center>
\"Google<\/center>
\n
Fonte: opennet.ru<\/a> <\/p>","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"excerpt":{"rendered":"

\u041a\u043e\u043c\u043f\u0430\u043d\u0438\u044f Google \u043e\u043f\u0443\u0431\u043b\u0438\u043a\u043e\u0432\u0430\u043b\u0430 \u043d\u043e\u0432\u043e\u0435 \u0441\u0435\u043c\u0435\u0439\u0441\u0442\u0432\u043e \u0431\u043e\u043b\u044c\u0448\u0438\u0445 \u044f\u0437\u044b\u043a\u043e\u0432\u044b\u0445 \u043c\u043e\u0434\u0435\u043b\u0435\u0439 Gemma 4, \u043e\u0441\u043d\u043e\u0432\u0430\u043d\u043d\u044b\u0445 \u043d\u0430 \u0442\u0435\u0445\u043d\u043e\u043b\u043e\u0433\u0438\u044f\u0445 \u043c\u043e\u0434\u0435\u043b\u0438 Gemini 3. Gemma 4 \u0440\u0430\u0441\u043f\u0440\u043e\u0441\u0442\u0440\u0430\u043d\u044f\u0435\u0442\u0441\u044f \u043f\u043e\u0434 \u043b\u0438\u0446\u0435\u043d\u0437\u0438\u0435\u0439 Apache \u0432 \u0432\u0430\u0440\u0438\u0430\u043d\u0442\u0430\u0445 \u0441 2.3, 4.5, 25.2 \u0438 30.7 \u043c\u0438\u043b\u043b\u0438\u0430\u0440\u0434\u0430\u043c\u0438 \u043f\u0430\u0440\u0430\u043c\u0435\u0442\u0440\u043e\u0432 (E2B, E4B, 31B \u0438 26B A4B). \u0412\u0430\u0440\u0438\u0430\u043d\u0442\u044b E2B \u0438 E4B \u043f\u043e\u0434\u0445\u043e\u0434\u044f\u0442 \u0434\u043b\u044f \u0438\u0441\u043f\u043e\u043b\u044c\u0437\u043e\u0432\u0430\u043d\u0438\u044f \u043d\u0430 \u043c\u043e\u0431\u0438\u043b\u044c\u043d\u044b\u0445 \u0443\u0441\u0442\u0440\u043e\u0439\u0441\u0442\u0432\u0430\u0445, \u0441\u0438\u0441\u0442\u0435\u043c\u0430\u0445 \u0438\u043d\u0442\u0435\u0440\u043d\u0435\u0442\u0430 \u0432\u0435\u0449\u0435\u0439 (IoT) \u0438 \u043f\u043b\u0430\u0442\u0430\u0445 \u0442\u0438\u043f\u0430 […]<\/p>\n","protected":false,"gt_translate_keys":[{"key":"rendered","format":"html"}]},"author":8,"featured_media":167259,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[702],"tags":[],"class_list":["post-167258","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-novosti-interneta"],"aioseo_notices":[],"aioseo_head":"\n\t\t\n\t\n\t\n\t\n\t\n\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\t\t\n\n","aioseo_head_json":{"title":"\ud83e\udd47Google ha rilasciato il modello AI open source Gemma 4, costruito sulla tecnologia Gemini 3 | ProHoster","description":"Google ha pubblicato una nuova famiglia di grandi modelli linguistici Gemma 4, basata sulla tecnologia del modello Gemini 3. Gemma 4 \u00e8 distribuito con licenza Apache in varianti da 2,3, 4,5, 25,2 e 30,7 miliardi di parametri (E2B, E4B, 31B e 26B A4B). Le varianti E2B e E4B sono adatte per l'uso su dispositivi mobili, sistemi Internet of Things (IoT) e schede tipo.","canonical_url":"https:\/\/prohoster.info\/it\/blog\/novosti-interneta\/google-vypustil-otkrytuyu-ai-model-gemma-4-postroennuyu-na-tehnologiyah-gemini-3","robots":"max-image-preview:large","keywords":"","webmasterTools":{"miscellaneous":""},"schema":null,"og:locale":"it_IT","og:site_name":"ProHoster | \u041a\u0443\u043f\u0438\u0442\u044c \u043d\u0430\u0434\u0435\u0436\u043d\u044b\u0439 \u0445\u043e\u0441\u0442\u0438\u043d\u0433 \u0434\u043b\u044f \u0441\u0430\u0439\u0442\u043e\u0432 \u0441 \u0437\u0430\u0449\u0438\u0442\u043e\u0439 \u043e\u0442 DDoS, VPS VDS \u0441\u0435\u0440\u0432\u0435\u0440\u044b","og:type":"article","og:title":"\ud83e\udd47Google \u0432\u044b\u043f\u0443\u0441\u0442\u0438\u043b \u043e\u0442\u043a\u0440\u044b\u0442\u0443\u044e AI-\u043c\u043e\u0434\u0435\u043b\u044c Gemma 4, \u043f\u043e\u0441\u0442\u0440\u043e\u0435\u043d\u043d\u0443\u044e \u043d\u0430 \u0442\u0435\u0445\u043d\u043e\u043b\u043e\u0433\u0438\u044f\u0445 Gemini 3 | ProHoster","og:description":"\u041a\u043e\u043c\u043f\u0430\u043d\u0438\u044f Google \u043e\u043f\u0443\u0431\u043b\u0438\u043a\u043e\u0432\u0430\u043b\u0430 \u043d\u043e\u0432\u043e\u0435 \u0441\u0435\u043c\u0435\u0439\u0441\u0442\u0432\u043e \u0431\u043e\u043b\u044c\u0448\u0438\u0445 \u044f\u0437\u044b\u043a\u043e\u0432\u044b\u0445 \u043c\u043e\u0434\u0435\u043b\u0435\u0439 Gemma 4, \u043e\u0441\u043d\u043e\u0432\u0430\u043d\u043d\u044b\u0445 \u043d\u0430 \u0442\u0435\u0445\u043d\u043e\u043b\u043e\u0433\u0438\u044f\u0445 \u043c\u043e\u0434\u0435\u043b\u0438 Gemini 3. Gemma 4 \u0440\u0430\u0441\u043f\u0440\u043e\u0441\u0442\u0440\u0430\u043d\u044f\u0435\u0442\u0441\u044f \u043f\u043e\u0434 \u043b\u0438\u0446\u0435\u043d\u0437\u0438\u0435\u0439 Apache \u0432 \u0432\u0430\u0440\u0438\u0430\u043d\u0442\u0430\u0445 \u0441 2.3, 4.5, 25.2 \u0438 30.7 \u043c\u0438\u043b\u043b\u0438\u0430\u0440\u0434\u0430\u043c\u0438 \u043f\u0430\u0440\u0430\u043c\u0435\u0442\u0440\u043e\u0432 (E2B, E4B, 31B \u0438 26B A4B). \u0412\u0430\u0440\u0438\u0430\u043d\u0442\u044b E2B \u0438 E4B \u043f\u043e\u0434\u0445\u043e\u0434\u044f\u0442 \u0434\u043b\u044f \u0438\u0441\u043f\u043e\u043b\u044c\u0437\u043e\u0432\u0430\u043d\u0438\u044f \u043d\u0430 \u043c\u043e\u0431\u0438\u043b\u044c\u043d\u044b\u0445 \u0443\u0441\u0442\u0440\u043e\u0439\u0441\u0442\u0432\u0430\u0445, \u0441\u0438\u0441\u0442\u0435\u043c\u0430\u0445 \u0438\u043d\u0442\u0435\u0440\u043d\u0435\u0442\u0430 \u0432\u0435\u0449\u0435\u0439 (IoT) \u0438 \u043f\u043b\u0430\u0442\u0430\u0445 \u0442\u0438\u043f\u0430","og:url":"https:\/\/prohoster.info\/it\/blog\/novosti-interneta\/google-vypustil-otkrytuyu-ai-model-gemma-4-postroennuyu-na-tehnologiyah-gemini-3","og:image":"https:\/\/prohoster.info\/wp-content\/uploads\/2021\/11\/logo-350.jpg","og:image:secure_url":"https:\/\/prohoster.info\/wp-content\/uploads\/2021\/11\/logo-350.jpg","og:image:width":350,"og:image:height":350,"article:published_time":"2026-04-03T03:11:56+00:00","article:modified_time":"2026-04-03T03:11:56+00:00","article:publisher":"https:\/\/www.facebook.com\/prohoster","article:author":"https:\/\/www.facebook.com\/prohoster"},"aioseo_meta_data":[],"gt_translate_keys":[{"key":"link","format":"url"}],"_links":{"self":[{"href":"https:\/\/prohoster.info\/it\/wp-json\/wp\/v2\/posts\/167258","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/prohoster.info\/it\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/prohoster.info\/it\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/prohoster.info\/it\/wp-json\/wp\/v2\/users\/8"}],"replies":[{"embeddable":true,"href":"https:\/\/prohoster.info\/it\/wp-json\/wp\/v2\/comments?post=167258"}],"version-history":[{"count":0,"href":"https:\/\/prohoster.info\/it\/wp-json\/wp\/v2\/posts\/167258\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/prohoster.info\/it\/wp-json\/wp\/v2\/media\/167259"}],"wp:attachment":[{"href":"https:\/\/prohoster.info\/it\/wp-json\/wp\/v2\/media?parent=167258"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/prohoster.info\/it\/wp-json\/wp\/v2\/categories?post=167258"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/prohoster.info\/it\/wp-json\/wp\/v2\/tags?post=167258"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}