Models with larger weights
Evaluate placing the model on one GPU while allowing for engine memory.
H200 SXM rental
141 GB on one GPU for model weights, KV cache and working buffers.
$4.232/ hour
Workloads for this GPU configuration
Long-context inference
Measure KV cache growth at your actual input and output lengths.
Concurrent requests
Balance request concurrency, latency and the 141 GB memory budget.
Software compatibility
- The GPU has 141 GB of HBM3e. Include runtime memory and working buffers in the total.
- HBM3e capacity helps fit data, but alone does not guarantee your target speed.
- Test context length and concurrent requests together.
Rent H200 SXM
Provide your model, software environment and rental period. We will confirm the rate, availability and connection method before the rental starts.
$4.232/ hour
About this configuration
How much memory does H200 SXM have?
The GPU has 141 GB of HBM3e. Include runtime memory and working buffers in the total.
Why 141 GB if the weights fit in 80 GB?
Additional memory may be needed for KV cache, concurrent requests or training state. Measure peak usage at your target load to understand whether the extra capacity helps.
How do I request this configuration?
This configuration is open for rental requests. Share your workload, preferred dates and duration. Hardware availability, pricing and access dates are confirmed individually; a request does not create a confirmed booking.
