Fit is always evaluated when model sizing is available. GPU rent, paid idle capacity, API crossover, and measured optimization appear only when their required evidence exists. When the API wins, the page says so.
Start with the computed memory answer and the recommended path forward. The breakdown below keeps the fit reproducible from cited inputs.
Model
meta-llama/Llama-3.3-70B-Instruct
GPU
NVIDIA H100 SXM 80 GB
Requests/day
50,000
GPU count
1
Quantization
FP16
Tokens/request
500 in, 500 out
Memory result
The model and KV cache need more VRAM than the selected GPU count provides. No serving cost is published for a configuration that cannot run.
Memory fit: 1 x AMD Instinct MI355X at 90% vLLM GPU memory utilization on one node. Attention head count was not resolved, so this is a memory topology, not a validated tensor-parallel size.
Computed actions and cost signals appear only when the result supports them.
FP16 weights use 141.1 GB and bring total memory to 201.2 GB; INT4, NVFP4, or MXFP4 would lower weights to 35.3 GB and total memory to 69.1 GB, which fits inside 72 GB.
Model
meta-llama/Llama-3.3-70B-Instruct
Hardware
1 x NVIDIA H100 SXM 80 GB, FP16
Weights and KV cache stay separate so the fit is reproducible.
10 sources, 7 caveats
Share this calculation
This result URL remains shareable. Badge embeds appear after memory fit and publication evidence are both complete.
