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
GambarTeksVideo
Harga in / out
₩123 / ₩597
1M
Context
262,14 rb
Dirilis
3 Apr 2026
Apps
Aplikasi yang paling banyak menggunakan model ini.
| Apps | tokens |
|---|---|
| Hermes Agent | 11,32 T |
| Zazen (Freebuff fork) | 3,16 T |
| Cline | 2,56 T |
| Kilo Code | 2,44 T |
| pi | 1,59 T |
| OpenClaw | 1,05 T |
| Codex | 938,81 M |
| DeepSeek Harness | 847,26 M |
| Portkey AI | 639,39 M |
| HighLevel | 375,1 M |
| ISEKAI ZERO | 364,56 M |
| Framer | 348,63 M |
| Hello Minds, powered by Ethoswarm | 316,97 M |
| ZCode | 279,83 M |
| Descript | 253,44 M |