Google: Gemma 4 26B A4B
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...
Modalities
ImageTextVideo
In / out price
₩123 / ₩597
1M
Context
2.62 लाख
रिलीज़
3 अप्रैल 2026
Apps
वे ऐप्स जो इस मॉडल का सबसे अधिक उपयोग करते हैं।
| Apps | tokens |
|---|---|
| Hermes Agent | 1.13 नील |
| Zazen (Freebuff fork) | 31.59 ख॰ |
| Cline | 25.57 ख॰ |
| Kilo Code | 24.42 ख॰ |
| pi | 15.88 ख॰ |
| OpenClaw | 10.46 ख॰ |
| Codex | 9.39 ख॰ |
| DeepSeek Harness | 8.47 ख॰ |
| Portkey AI | 6.39 ख॰ |
| HighLevel | 3.75 ख॰ |
| ISEKAI ZERO | 3.65 ख॰ |
| Framer | 3.49 ख॰ |
| Hello Minds, powered by Ethoswarm | 3.17 ख॰ |
| ZCode | 2.8 ख॰ |
| Descript | 2.53 ख॰ |