Skip to content
Updated minutes ago
other

Groq LPU

other · other · 230 GB SRAM · 300W TDP

VRAM

230 GB

BF16 TFLOPS

188

Bandwidth

80000 GB/s

From

$0.00/hr

Calculate ROI with this GPU →

Spec Sheet

VRAM230 GB SRAM
Memory Bandwidth80000 GB/s
BF16 TFLOPS188
FP16 TFLOPS188
FP8 TFLOPS376
INT8 TOPS750
TDP300W
InterconnectPCIE
Max per Node8
PCIe Gen5
Tensor CoresNo

Pricing by Provider

ProviderOn-DemandReservedSpotBadge
groq$0.00/hr--Cheapest

Compatible Models (280)

Training Capabilities

Estimated GPU count for full fine-tuning (AdamW, BF16) and QLoRA

Model SizeFull Fine-TuneQLoRA
7B model1 GPU1 GPU
13B model2 GPUs1 GPU
70B model6 GPUs1 GPU

Energy Efficiency

Estimated tokens/second per Watt for popular models

Mistral 7B
36.53 t/s/WFP8
Qwen 2.5 7B
35.09 t/s/WFP8
Llama 3.1 8B
33.21 t/s/WFP8
DeepSeek V3
7.21 t/s/WFP8
Llama 3.1 70B
3.78 t/s/WFP8
Qwen 2.5 72B
3.67 t/s/WFP8

Similar GPUs

GPUVRAMBF16 TFLOPSBW (GB/s)From
Trainium296 GB7563200$1.95/hr
Cloud AI 10032 GB150134$0.00/hr
Instinct MI325X256 GB13076000$2.49/hr
B100 SXM192 GB17508000$4.50/hr
GB200 NVL72 (per GPU)192 GB22508000$6.50/hr

Methodology Note

Performance estimates for the Groq LPUare based on InferenceBench's roofline performance model with CUDA kernel-level optimization including FlashAttention v2 and PagedAttention. Memory calculations account for model weights (230 GB SRAM available), KV-cache allocation, and activation memory. Throughput predictions use the Groq LPU's rated 80000 GB/s memory bandwidth and 188 BF16 TFLOPS compute capacity as roofline ceilings, with empirical correction factors per GPU architecture (other). See our full methodology.

Frequently Asked Questions

How many AI models can run on Groq LPU?

The Groq LPU can run 280 AI models from our database within a single node. Compatible models range across various parameter sizes depending on the quantization precision (BF16, FP8, INT4). Smaller models fit on a single GPU while larger models may require multi-GPU setups up to 8x Groq LPU.

What is the Groq LPU inference throughput?

The Groq LPU delivers 188 BF16 TFLOPS and 376 FP8 TFLOPS with 80000 GB/s memory bandwidth. Actual inference throughput (tokens/sec) depends on the model size, precision, and batch size. Use our calculator for model-specific throughput estimates.

How much does Groq LPU cost per hour?

The Groq LPU is available starting from $0.00/hour via groq. Prices vary by provider and pricing tier (on-demand, reserved, spot). Compare pricing across all providers in the table above.