DeepSeek R1
Partner VerifiedPublisher: DeepSeek AI · Category: Reasoning
Parameters
32B Distill
Context Window
128K
Default Size
20.5 GB
Downloads
520,400
Open-weights reasoning model with advanced chain-of-thought processing and state-of-the-art math/coding performance.
PULL MODEL
eq pull deepseek-r1
RUN INFERENCE
eq run deepseek-r1 "Solve differential equation y'' + y = 0"
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.