Gemma 3n E4B Instructed

Fiche modèle · Google · gemma-3n-e4b-it

Gemma 3n is a multimodal model designed to run locally on hardware, supporting image, text, audio, and video inputs. It features a language decoder, audio encoder, and vision encoder, and is available in two sizes: E2B and E4B. The model is optimized for memory efficiency, allowing it to run on devices with limited GPU RAM. Gemma is a family of lightweight, state-of-the-art open models from Google, built from the same research and technology used to create the Gemini models. Gemma models are well-suited for a variety of content understanding tasks, including question answering, summarization, and reasoning. Their relatively small size makes it possible to deploy them in environments with limited resources such as laptops, desktops or your own cloud infrastructure, democratizing access to state of the art AI models and helping foster innovation for everyone. Gemma 3n models are designed for efficient execution on low-resource devices. They are capable of multimodal input, handling text, image, video, and audio input, and generating text outputs, with open weights for instruction-tuned variants. These models were trained with data in over 140 spoken languages.

Identité

OrganisationGoogle
Pays—
Famille—
Typellm
Date de sortie2025-06-26
Licenceproprietary
Licence commercialenon
Licence open sourcenon
Poids ouvertsnon
Auto-hébergement (DGX Spark)non — poids fermés
Modalitésimage|text
Fenêtre de contexte—
KV cache / token (Q8)—
Paramètres totaux8 B
Paramètres actifs (session)4 B
Source des paramètres actifsweb : https://ai.google.dev/gemma/docs/gemma-3n/model_card
Curation des paramètres actifs2026-10-19
Prix par session—
Tokens / tâche (Data Players)1.2 M
Tokens / tâche (Terminal-Bench)83.3 M (est.)
Ratio Terminal-Bench / Data Players69.4×
Coût / tâche (Terminal-Bench)—
Tokens d'entraînement11 000 000 000 000
Coupure de connaissance2024-06-01
Créé le2026-05-07T16:53:29.984057Z
Mis à jour le2026-11-04T06:04:46.265799Z
SourceLLM Stats
URLhttps://llm-stats.com/models/gemma-3n-e4b-it

Prix par fournisseur

Toutes sources confondues : catalogue llm-stats, routeurs EUrouter/Cortecs et catalogues directs (OVHcloud, Scaleway, Infomaniak). Les prix sont convertis en USD pour la comparaison ; le prix publié et le taux BCE utilisé sont dans l'infobulle. Les prix inconnus restent indiqués « — ».

Aucune donnée.

Souveraineté qualifiée par offre (modèle × fournisseur) ; survolez le badge pour les critères.

Souveraineté

Scores par catégorie (top_scores)

CatégorieScore
language0.50
legal0.50
math0.40
physics0.20
reasoning0.40
finance0.50
general0.40
healthcare0.50
biology0.20
chemistry0.20
code0.40

Scores par benchmark

BenchmarkCatégorieScoreMaxNormalisé (API)VérifiéAuto-déclaréDaté du
aime-2025math0.121.000.12nonoui2026-11-04T06:04:46.276654Z
codegolf-v2.2code0.171.000.17nonoui2026-11-04T06:04:46.280087Z
eclekticlanguage0.191.000.19nonoui2026-11-04T06:04:46.284799Z
global-mmlugeneral0.601.000.60nonoui2026-11-04T06:04:46.287832Z
global-mmlu-litegeneral0.651.000.65nonoui2026-11-04T06:04:46.291086Z
gpqareasoning0.241.000.24nonoui2026-11-04T06:04:46.294948Z
hiddenmathmath0.381.000.38nonoui2026-11-04T06:04:46.299155Z
humanevalreasoning0.751.000.75nonoui2026-11-04T06:04:46.303405Z
includegeneral0.571.000.57nonoui2026-11-04T06:04:46.312964Z
livecodebenchreasoning0.131.000.13nonoui2026-11-04T06:04:46.323121Z
livecodebench-v5reasoning0.261.000.26nonoui2026-11-04T06:04:46.327792Z
mbppreasoning0.64100.000.64nonoui2026-11-04T06:04:46.333415Z
mgsmmath0.671.000.67nonoui2026-11-04T06:04:46.339938Z
mmlugeneral0.651.000.65nonoui2026-11-04T06:04:46.347130Z
mmlu-prolanguage0.511.000.51nonoui2026-11-04T06:04:46.350838Z
mmlu-proxlanguage0.201.000.20nonoui2026-11-04T06:04:46.360949Z
openai-mmlugeneral0.361.000.36nonoui2026-11-04T06:04:46.365009Z
wmt24++language0.501.000.50nonoui2026-11-04T06:04:46.369123Z

Classements

CatégorieRangScoreNote conservatriceBenchmarks évaluésMéthodeDaté duProvenance
code257—-0.173trueskill2026-10-19model_history

Lecture : Note conservatrice = borne basse TrueSkill (niveau estimé − marge d’incertitude) ; Benchmarks évalués = nombre de benchmarks de la catégorie pris en compte. À performances égales, un modèle évalué sur moins de benchmarks obtient une note plus basse : la note ne se déduit pas d’un seul score. Provenance : rankings = top 50 de /rankings ; model_history = historique du modèle (/models/{id}/history), utilisé pour les rangs au-delà du top 50.

Données

Dates de curation (récupération depuis l'API) : models 2026-11-04 · benchmarks 2026-11-04 · scores 2026-11-04 · updates 2026-11-04 · account 2026-11-04 · model_history 2026-11-04 · rankings 2026-11-04