Google: Gemma 4 26B A4B
Fiabilité
20%
google/gemma-4-26b-a4b-it
Gemma 4 26B A4B IT is an instruction-tuned Mixture-of-Experts (MoE) model from Google DeepMind. Despite 25.2B total parameters, only 3.8B activate per token during inference — delivering near-31B quality at a fraction of the compute cost. Supports multimodal input including text, images, and video (...
🧠 Intelligence & Données
Knowledge Cutoff:
Inconnue
Tokenizer:
Gemma
Moderation:
✅ Non
📅 Cycle de vie
Ajouté le:
03/04/2026
Spécifications
-
Provider & Modalité google text+image+video->text
-
Fenêtre de contexte 262,144 tokens
-
Max Output Tokens 262,144
-
Support des Outils (Tools) ✔️ Fonction Calling
🔍 Modèles similaires
| Modèle | Provider | Input | Output | Contexte | |
|---|---|---|---|---|---|
|
Google: Gemma 4 31B
google/gemma-4-31b-it
|
$0.1400 | $0.4000 | 262,144 | → | |
|
Z.ai: GLM 5V Turbo
z-ai/glm-5v-turbo
|
z-ai | $1.2000 | $4.0000 | 202,752 | → |
|
Arcee AI: Trinity Large Thinking
arcee-ai/trinity-large-th...
|
arcee-ai | $0.2200 | $0.8500 | 262,144 | → |
|
Google: Lyria 3 Pro Preview
google/lyria-3-pro-previe...
|
$0.0000 | $0.0000 | 1,048,576 | → | |
|
Google: Lyria 3 Clip Preview
google/lyria-3-clip-previ...
|
$0.0000 | $0.0000 | 1,048,576 | → |