Rent NVIDIA H200 SXM in the Cloud

According to our July 13, 2026 snapshot, NVIDIA H200 SXM cloud instances begins at $2.05 per hour, with Vast.ai offering the lowest-cost on-demand rate across the 3 providers we track. Spot (preemptible) pricing on supporting clouds starts from $3.59 per hour — a discount of roughly up to 60% off on-demand, which makes short fine-tuning runs and checkpointable jobs substantially cheaper.

Reserved / commitment pricing cuts rates further by up to 40%, at the cost of locking in hours or months of capacity up front. For most AI teams, the sweet spot is spot for overnight fine-tuning and on-demand for interactive notebooks or low-latency real-time serving.

Get NVIDIA H200 SXM cloud access from Vast.ai, DigitalOcean, or RunPod. Live pricing, fast provisioning, competitive rates.

The NVIDIA H200 SXM datasheet shows a Hopper-class card with 141 GB of HBM3e memory, 4,800 GB/s memory bandwidth, 990 TFLOPS of FP16 compute, 67 TFLOPS of FP32 compute, and a 700W thermal envelope. It reached market in 2024.

Those specs make it capable of handling the full modern AI stack: pre-model training moderate-sized models, fine-tuning most LLMs, running production real-time serving at useful batch sizes, and accelerating diffusion / image generation workloads.

Spin up NVIDIA H200 SXM on Vast.ai for $2.05/hr, with DigitalOcean and RunPod as strong alternatives — compare and deploy in minutes.

NVIDIA H200 SXM is a Hopper card offering 990 FP16 TFLOPS and 67 FP32 TFLOPS alongside 4,800 GB/s of memory bandwidth. That's enough compute to handle modern model training and real-time serving workloads at serious scale.

Benchmarks show NVIDIA H200 SXM performs particularly well on transformer-style models where tensor cores are saturated by large MatMuls. Diffusion models, speech, and vision workloads also see strong speedups versus older generations. For latency-sensitive production real-time serving, NVIDIA H200 SXM usually hits target token-per-second rates on large language models well above the 30-50 tok/s threshold most products aim for.

Spin up NVIDIA H200 SXM on Vast.ai for $2.05/hr, with DigitalOcean and RunPod as strong alternatives — compare and deploy in minutes.

The short list of providers carrying NVIDIA H200 SXM: DigitalOcean, RunPod, Vast.ai. We track 3 in total. The best-priced on-demand rate is at Vast.ai ($2.05 per hour), and the best-priced spot rate is at RunPod ($3.59 per hour).

When evaluating any of them, don't just compare price — check regional coverage, compliance (SOC 2, HIPAA, GDPR), multi-GPU interconnect, and the actual realised price once egress fees are factored in. A cheap hourly rate can evaporate fast once you move the outputs of a training run off the provider's network.

Spin up NVIDIA H200 SXM on Vast.ai for $2.05/hr, with DigitalOcean and RunPod as strong alternatives — compare and deploy in minutes.

NVIDIA H200 SXM is best for workloads where its 141 GB VRAM and Hopper tensor cores are well-matched: Large model training, inference for 70B+ parameter models.

If your workload needs significantly more memory (e.g., training frontier-scale models from scratch), NVIDIA H200 SXM is undersized and you'd want an H100/H200/B200 class card. If your workload needs less (e.g., small-scale serving on 7B-parameter models), cheaper cards like L4 or RTX 4090 may be more cost-efficient. For the middle band, NVIDIA H200 SXM is usually the sensible pick.

Rent NVIDIA H200 SXM today from Vast.ai, DigitalOcean, or RunPod — compare live pricing and spin up in minutes.