Skip to content

Instinct MI350X vs Instinct MI300X

Complete side-by-side comparison of specs, performance, memory, power efficiency, and pricing.

AMD

Instinct MI350X

93

Spec Wins

AMD

Instinct MI300X

65

Detailed Specifications

SpecInstinct MI350XInstinct MI300X
ArchitectureCDNA 4 CDNA 3
Memory288GB HBM3e 192GB HBM3
Memory Bandwidth8,000 GB/s 5,300 GB/s
FP16 TFLOPS2,307 1,307
FP8 TFLOPS4,614 2,614
BF16 TFLOPS2,307 1,307
INT8 TOPS4,614 2,614
TDP1000W 750W
InterconnectInfinity Fabric 4.0 (896 GB/s) (896 GB/s) Infinity Fabric (896 GB/s) (896 GB/s)
Perf Score93 65
EcosystemROCM ROCM
Est. Price$25,000 $15,000

Instinct MI350X — Best For

LLM TrainingAI InferenceHPC

Instinct MI300X — Best For

Large ModelsCost-Effective Training

Who Should Choose Each GPU?

Choose Instinct MI350X if you…

  • Need more VRAM (288GB vs 192GB) for large model inference
  • Prioritize raw FP8 throughput (4,614 vs 2,614 TFLOPS)
  • Running LLM Training workloads
  • Running AI Inference workloads
  • Running HPC workloads

Choose Instinct MI300X if you…

  • Have power-constrained data centers (750W vs 1000W TDP)
  • Working with a tighter CapEx budget (lower list price)
  • Running Large Models workloads
  • Running Cost-Effective Training workloads

Verdict

The Instinct MI350X and Instinct MI300X target different priorities. The Instinct MI350X's 288GB of HBM3e gives it a clear edge for large-model inference where fitting the full model in VRAM eliminates quantization overhead. For training throughput, the Instinct MI350X's 4,614 FP8 TFLOPS outpaces the Instinct MI300X's 2,614 TFLOPS. Both GPUs use ROCM, so ecosystem switching cost is not a factor. Use our TCO Calculator to model the full 3-year cost difference for your specific utilization and power costs.

Instinct MI350X vs Instinct MI300X: Common Questions

Which is faster, Instinct MI350X or Instinct MI300X?+

In FP8 throughput, the Instinct MI350X leads with 4,614 TFLOPS vs 2,614 TFLOPS. For LLM inference, memory capacity and bandwidth often matter more than raw TFLOPS — the Instinct MI350X has more VRAM (288GB).

Is Instinct MI350X or Instinct MI300X better for LLM training?+

For LLM training at scale, the Instinct MI350X has higher raw throughput. However, the choice also depends on your software stack: Instinct MI300X offers CUDA compatibility with the widest framework support (PyTorch, JAX, TensorRT).

What is the price difference between Instinct MI350X and Instinct MI300X?+

The Instinct MI350X is estimated at $25,000 per unit and the Instinct MI300X at $15,000. Actual pricing varies by vendor, volume, and configuration. Check our Buy page for current reseller pricing.

Which GPU is more power efficient, Instinct MI350X or Instinct MI300X?+

The Instinct MI300X has a lower TDP (750W vs 1000W). Performance-per-watt depends on your workload — for FP8 inference, divide TFLOPS by TDP: Instinct MI350X = 4.6 TFLOPS/W vs Instinct MI300X = 3.5 TFLOPS/W.

Ask AI Advisor