Inferbase
Beta
AMD Instinct MI250
Datacenter
The MI250 delivers dual-GCD CDNA 2 performance with 128GB HBM2e for demanding workloads. Optimized for high-performance computing and large-scale AI training.
VRAM
128 GB
Memory
HBM2e
Bandwidth
3277 GB/s
TDP
500W
Suitable Workloads
Large Language Models
Training and inference for models like GPT-4, Llama 70B+
High-Throughput Inference
Optimized for batched inference workloads
Enterprise Deployment
Designed for 24/7 datacenter operations
Key Highlights
- Built on CDNA 2 architecture
- 832 AI accelerator cores for tensor operations
- 13,312 compute cores for parallel processing
- OAM form factor
Memory
VRAM128 GB
Memory TypeHBM2e
Memory Bandwidth3277 GB/s
Compute Performance
FP32 (Single Precision)45.3 TFLOPS
FP16 (Half Precision)362 TFLOPS
BF16362 TFLOPS
INT8362 TOPS
Architecture
ArchitectureCDNA 2
Compute Cores13,312
AI Accelerators832
Release Year2021
Power & Physical
TDP500 W
Max Power500 W
Form FactorOAM
PCIe GenerationGen4
Notes
Dual-GCD design for HPC and AI training workloads.
Models That May Fit on AMD Instinct MI250
Estimates based on INT8 quantization. Actual fit depends on framework and batch size.
Related Tools
Added Jan 27, 2026
Last updated: Jan 27, 2026
Build Your AI Stack with Confidence
Explore models, compare pricing and benchmarks, and right-size your infrastructure — all in one place.