google/gemma-4-26B-A4B-it

Gemma 4 26B A4B-it: A Multimodal MoE for Advanced Reasoning and Coding

This model is part of the Gemma 4 family, developed by Google DeepMind, offering a 25.2 billion total parameter Mixture-of-Experts (MoE) architecture with 3.8 billion active parameters. This design allows for faster inference compared to dense models of similar total size. It is an instruction-tuned variant capable of processing both text and image inputs, with a substantial 256K token context window.

Key Capabilities

• Multimodal Input: Processes text and images, with variable aspect ratio and resolution support.
• Reasoning: Designed as a highly capable reasoner with configurable thinking modes.
• Extended Context: Supports a 256K token context window for complex, long-context tasks.
• Efficient Architecture: Utilizes a Mixture-of-Experts (MoE) design for optimized inference speed.
• Enhanced Coding & Agentic Capabilities: Achieves notable improvements in coding benchmarks and includes native function-calling support for autonomous agents.
• Native System Prompt Support: Facilitates more structured and controllable conversations.

Good For

• Reasoning-intensive applications: Leveraging its advanced reasoning capabilities and configurable thinking modes.
• Agentic workflows: Utilizing native function-calling support for building autonomous agents.
• Coding tasks: Excelling in code generation, completion, and correction.
• Multimodal understanding: Processing and interpreting interleaved text and image inputs for diverse applications like document parsing, UI understanding, and OCR.
• Deployment on consumer GPUs and workstations: Offering a balance of performance and efficiency due to its MoE architecture.