Gemma 4 26B-A4B
Released 2026-04 · chat · 256K tokens · 5 benchmarks · Open weight
Editorial notes
Variante MoE: 26B totales / 4B activos por token. Multimodal (texto, imagen, video). Mejor relacion performance/active-params del catalogo Gemma 4. Codeforces ELO 1718.
Spec sheet
- Empresa
- Google DeepMind
- Pais
- US
- Tipo
- chat
- Release
- 2026-04
- Context
- 256K tokens
- Params total
- 25.2B
- Params active (MoE)
- 3.8B
- Licencia
- apache-2.0
- Quants
- BF16, Q8_0, Q5_K_M, Q4_K_M, Q3_K_M
- Slug
- gemma-4-26b-a4b
Quick install
2 toolsBenchmarks (5)
Reasoning 3
Cite this model
BibTeX · APA
BibTeX
@misc{frontier-gemma-4-26b-a4b,
title = {Gemma 4 26B-A4B},
author = {{Google DeepMind}},
year = {2026},
note = {Frontier Benchmarks AI atlas. Accessed 2026-06-17},
url = {https://frontierbenchmarks.com/models/gemma-4-26b-a4b}
} APA
Google DeepMind (2026). Gemma 4 26B-A4B [Large language model]. Frontier Benchmarks AI. Retrieved 2026-06-17, from https://frontierbenchmarks.com/models/gemma-4-26b-a4b
Citation refleja la pagina del atlas, no el paper original del modelo. Para el paper, ve a la seccion "Recursos" arriba.