Gemma 4 Models
Compare all Gemma 4 model sizes, parameters, context windows, and hardware requirements to choose the right checkpoint.
What are Gemma 4 Models?
Gemma 4 ships in four sizes — E2B, E4B, 26B A4B (MoE), and 31B Dense. Each is tuned for different hardware and use cases, from mobile devices to high-end workstations.
Why compare Gemma 4 models?
Choose the Right Size
Select the model checkpoint that fits your hardware constraints and quality requirements
Understand Dense vs MoE
Learn why the 26B MoE runs faster than you'd expect compared to a 26B Dense model
Plan Your Deployment
Know exact memory and storage requirements before downloading large checkpoints
Featured & Essential
Gemma 4 Size
Explore the different Gemma 4 size options, from edge-ready models to powerful workstation tiers. Compare parameters, hardware requirements, and multimodal features.
Gemma 4 Multimodal
Explore the groundbreaking capabilities of Gemma 4 multimodal models. Learn about the 26B and 31B architectures, gaming performance, and local deployment tips.
All Gemma 4 Model Guides
Gemma 4 26b a4b
Explore the power of the gemma 4 26b a4b model. Learn about its Mixture of Experts architecture, performance benchmarks against GPT-5.4, and logic optimization tips.
Gemma 4 26B Guide
A comprehensive guide to the Gemma 4 26B Mixture of Experts model. Learn about its architecture, local performance, and agentic capabilities in 2026.
Gemma 4 31B
Explore the groundbreaking Gemma 4 31B model. Learn about its 256k context window, multimodal gaming capabilities, and local deployment performance.
Gemma 4 E2B
Explore the capabilities of Gemma 4 E2B, Google's latest edge-optimized AI model. Learn about its native multimodality, thinking features, and Apache 2.0 license.
Gemma 4 E4B
Explore the Gemma 4 E4B model, Google's latest breakthrough in edge AI. Learn about its effective parameters, PLE architecture, and agentic capabilities for 2026.
Gemma 4 Models
Explore the full capabilities of the Gemma 4 models. Learn about the 26B MoE and 31B Dense variants, their gaming applications, and performance benchmarks.