Back to catalog

VRAM Calculator

Estimate GPU memory requirements for inference. Adjust parameters to see real-time VRAM breakdown.

Model Parameters

VRAM Estimate

Total VRAM Required
163.68 GB
at FP16 / BF16 · 8,192 ctx · batch 1
0 GB35 GB70 GB105 GB140 GBModelWeightsKV CacheActivations
Weights
140 GB
KV Cache
2.68 GB
Activations
21 GB

Quantization Comparison

PrecisionWeightsKV CacheActivationsTotal
FP16 / BF16140 GB2.68 GB21 GB163.68 GB
INT870 GB1.34 GB10.5 GB81.84 GB
INT435 GB0.67 GB5.25 GB40.92 GB

Estimates account for model weights, KV cache (2 × layers × kv_heads × head_dim × seq_len × batch × bytes), and activations (~15% of weights). Actual usage varies by framework, kernel efficiency, and fragmentation.