Llama 3.3 70B
Partner VerifiedPublisher: Meta AI · Category: Chat
Parameters
70B
Context Window
128K
Default Size
42.5 GB
Downloads
840,100
Meta flagship 70-billion parameter open model offering state-of-the-art general intelligence and instruction adherence.
PULL MODEL
eq pull llama3.3:70b
RUN INFERENCE
eq run llama3.3:70b "Draft a technical architecture specification"
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.