Gemma 3 27B
Partner VerifiedPublisher: Google · Category: Multimodal
Parameters
27B
Context Window
128K
Default Size
16.8 GB
Downloads
325,600
Google next-generation open weight model built from the same research as Gemini 2 models.
PULL MODEL
eq pull gemma3:27b
RUN INFERENCE
eq run gemma3:27b "Analyze system metrics log"
Attention Architecture
Utilizes Grouped-Query Attention (GQA) with 16-token fixed Paged KV Cache block allocation for zero-fragmentation memory residency.
ONNX & EQC Compatible
Compiles directly through the EQC toolchain into standalone .eqx binary packages with fused SwiGLU kernels.
Multi-Hardware Support
Auto-detects CUDA RTX/A100/H100, Apple Silicon Metal Performance Shaders, or AVX-512 CPU execution backends.