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...
Modalitäten
BildTextVideo
In / out Preis
123 ₩ / 597 ₩
1M
Kontext
262.144
Veröffentlicht
03.04.2026
Apps
Apps, die dieses Modell am häufigsten nutzen.
| Apps | tokens |
|---|---|
| Hermes Agent | 11,32 Bio. |
| Zazen (Freebuff fork) | 3,16 Bio. |
| Cline | 2,56 Bio. |
| Kilo Code | 2,44 Bio. |
| pi | 1,59 Bio. |
| OpenClaw | 1,05 Bio. |
| Codex | 938,81 Mrd. |
| DeepSeek Harness | 847,26 Mrd. |
| Portkey AI | 639,39 Mrd. |
| HighLevel | 375,1 Mrd. |
| ISEKAI ZERO | 364,56 Mrd. |
| Framer | 348,63 Mrd. |
| Hello Minds, powered by Ethoswarm | 316,97 Mrd. |
| ZCode | 279,83 Mrd. |
| Descript | 253,44 Mrd. |