Updated minutes ago
Instinct MI250X
amd · cdna2 · 128 GB HBM2e · 560W TDP
VRAM
128 GB
BF16 TFLOPS
383
Bandwidth
3277 GB/s
From
$0.79/hr
Spec Sheet
VRAM128 GB HBM2e
Memory Bandwidth3277 GB/s
BF16 TFLOPS383
FP16 TFLOPS383
FP8 TFLOPS383
INT8 TOPS383
TDP560W
InterconnectINFINITY-FABRIC
Max per Node8
PCIe Gen4
Tensor CoresNo
Pricing by Provider
| Provider | On-Demand | Reserved | Spot | Badge |
|---|---|---|---|---|
| fluidstack | $1.09/hr | - | $0.79/hr | Cheapest |
| vast_ai | $1.20/hr | - | $0.85/hr |
Compatible Models (249)
Single GPU (215 models)
Command R+104B FP8Yi-Large102.6B FP8YaLM 100B100B FP8Llama 3.2 90B Vision90B FP8Llama 3.2 90B Vision Instruct88.8B FP8Qwen 2.5 72B72.7B FP8Qwen 2.5 Math 72B72.7B FP8Qwen 2.5 VL 72B72.7B FP8Dolphin 2.9 72B72B FP8DeepSeek R1 Distill 70B70.6B FP8Llama 3 70B 1M Context70.6B FP8Llama 3 70B70.6B FP8Llama 3.1 70B70.6B FP8Llama 3.3 70B70.6B FP8Hermes 3 70B70.6B FP8HelpSteer2 Llama 3.1 70B70.6B FP8Llama 3.1 Nemotron 70B Instruct70.6B FP8Llama 3.1 Nemotron 70B Reward70.6B FP8Nemotron 70B70.6B FP8Llama 3.1 70B Turbo70.6B FP8+195 more
Multi-GPU (34 models)
Training Capabilities
Estimated GPU count for full fine-tuning (AdamW, BF16) and QLoRA
| Model Size | Full Fine-Tune | QLoRA |
|---|---|---|
| 7B model | 2 GPUs | 1 GPU |
| 13B model | 2 GPUs | 1 GPU |
| 70B model | 11 GPUs | 1 GPU |
Energy Efficiency
Estimated tokens/second per Watt for popular models
Mistral 7B
0.80 t/s/WFP8
Qwen 2.5 7B
0.77 t/s/WFP8
Llama 3.1 8B
0.73 t/s/WFP8
DeepSeek V3
0.16 t/s/WFP8
Llama 3.1 70B
0.08 t/s/WFP8
Qwen 2.5 72B
0.08 t/s/WFP8
Similar GPUs
| GPU | VRAM | BF16 TFLOPS | BW (GB/s) | From |
|---|---|---|---|---|
| Instinct MI210 | 64 GB | 181 | 1638 | $0.99/hr |
| Gaudi 3 | 128 GB | 1835 | 3700 | $2.30/hr |
| Gaudi 3 HL-325L | 128 GB | 1835 | 3700 | $2.65/hr |
| H200 SXM | 141 GB | 990 | 4800 | $2.69/hr |
| H20 | 96 GB | 148 | 4000 | $0.99/hr |