gemma4-26B-A4B
This model is a fine-tuned version of the Gemma 4 26B architecture. It has been optimized using the A4B (Adapter-for-Business/Balance) methodology to provide a high-efficiency balance between reasoning capabilities and conversational fluidity. With 26 billion parameters, it sits in the "sweet spot" for complex tasks that require more nuance than 7B models but less compute than ultra-large models.
About Model
Gemma 4 26B A4B is a state-of-the-art Mixture-of-Experts (MoE) open model developed by Google DeepMind. While it contains 25.2B total parameters, it only activates approximately 3.8B parameters per token during inference. This allows the model to deliver the reasoning and quality of a ~31B dense model while maintaining the speed and low compute requirements of a much smaller system.
gemma4-26B-A4B
Metadata
Apache 2.0
Image Feature Extraction
Transformers
Open
Science, Technology and Research
20/04/26 07:24:51
48.10 GB
.gitattributes ( 1.53 KB )
To preview this file, you need to be a registered user. Please complete the registration process to gain access and continue viewing the content.
Activity Overview
0
0 
48.10 GB 
8
Tags
- advanced-reasoning
- Image
- Intelligent Document Processing
License Control
Apache 2.0
Version Control
Version 1(48.10 GB)
- admin·11 day(s) ago
.gitattributes
config.json
generation_config.json
model-00001-of-00002.safetensors
model-00002-of-00002.safetensors
model.safetensors.index.json
processor_config.json
README.md
tokenizer_config.json
More Models from DAKSH SOLUTIONS AND SERVICES
- 0
- 0
- 48.10 GB
8
© 2026 - Copyright AIKosh. All rights reserved. This portal is developed by National e-Governance Division for AIKosh mission.