NVIDIA H200
Verified offers20
Price floor$2.51per GPU-hour
Median$3.12per GPU-hour
Choose H200 when 141 GB is sufficient and maturity, current availability, or a lower infrastructure rate outweighs Blackwell uplift.
Review NVIDIA H200 pricesNVIDIA H200 vs B200 prices
Compare mature H200 capacity with newer B200 systems using live public offers and the complete delivered-node specification.
Bottom line
H200 is the maturity and availability baseline. B200 earns its premium when Blackwell performance, 180 GB of memory, higher memory bandwidth, or protected multi-GPU peer traffic produces a measurable workload advantage.
Marketplace data checked 2026-09-08. Specifications vary by form factor and delivered system.
Compare the buying case
Separate marketplace signals for each GPU, followed by the technical differences that affect workload cost.
NVIDIA B200
Verified offers9
Price floor$3.50per GPU-hour
Median$5.99per GPU-hour
Choose B200 when benchmarked Blackwell throughput, memory headroom, or protected peer traffic justifies the commercial and deployment premium.
Review NVIDIA B200 prices| Decision factor | NVIDIA H200 | NVIDIA B200 |
|---|---|---|
| Architecture | Hopper | Blackwell |
| GPU memory | 141 GB HBM3e | 180 GB HBM3e |
| Memory bandwidth | 4.8 TB/s | Up to 8 TB/s |
| Typical topology | HGX H200 partner systems with 4 or 8 GPUs; H200 NVL options also exist | HGX B200 systems use 8 GPUs connected through NVLink and NVSwitch |
| Marketplace availability | More established verified supply in the current comparison | Newer supply with fewer verified offers and greater delivery variance |
| Workload fit | Large-memory Hopper training and inference with a mature software and operating path | High-scale training, long-context or large-model inference, and sensitive multi-GPU workloads |
Reference specifications: NVIDIA H200 specifications ↗ · NVIDIA HGX platform specifications ↗
Current public offers
Filter the source-linked records by GPU. Prices retain their published region, commitment, and node context.
| GPU, provider and node | Region | Term | Normalized price | Source |
|---|---|---|---|---|
| H200FluidCore8x H200 141GB bare-metal node | Not published | hourly | $2.51/GPU-hr20.08 | Provider source ↗Checked 2026-08-23 |
| H200Clore.ai8x H200 Colorado bare-metal node | Colorado, USA | 360 days | $2.59/GPU-hr179020.80 | Provider source ↗Checked 2026-08-23 |
| H200Leaseweb1x H200 (HP DL385 G11) | London | term not stated | $2.80/GPU-hr£1,503.02/mo | Provider source ↗Checked 2026-08-20 |
| H200Taiga Cloud8x H200 SXM | Chester, UK | hourly | $3.00/GPU-hrFrom $3.00/GPU/hr (bare-metal tier) | Provider source ↗Checked 2026-09-04 |
| H200Qubrid AI8x H200 141GB bare-metal server | Not published | 12 months | $3.00/GPU-hr210240.00 | Provider source ↗Checked 2026-08-23 |
| H200DataPacket8x H200 NVL | Not published | 3–6-mo prepay typ. | $3.06/GPU-hr15368 | Provider source ↗Checked 2026-09-08 |
| H200Gcore8x H200 | Not published | hourly | $3.17/GPU-hr21.84 | Provider source ↗Checked 2026-09-04 |
| B200Vultr8x B200 192GB (HGX B200) | Not published | term not published | $3.50/GPU-hr$3.500/GPU/hr (pre-paid contract) | Provider source ↗Checked 2026-08-20 |
| B200Qubrid AI8x B200 180GB bare-metal server | Not published | 12 months | $4.25/GPU-hr297840.00 | Provider source ↗Checked 2026-08-23 |
| H200Hostrunway2x H200 (GPU Montreal 9) | Montreal | no lock-in | $4.45/GPU-hr$6,490/mo | Provider source ↗Checked 2026-08-20 |
| H200Cirrascale8x H200 | US | 1-mo min | $4.54/GPU-hr$26,499/mo (annual commit $21,199/mo) | Provider source ↗Checked 2026-08-20 |
| H200Hostrunway1x H200 dedicated GPU server (dual EPYC 9224) | Montreal | no lock-in; 15% discount at 12 months | $4.89/GPU-hr3572.00 | Provider source ↗Checked 2026-08-29 |
| B200Cirrascale8x B200 | US | 1-mo min | $5.99/GPU-hr$34,999/mo (annual commit $27,999/mo) | Provider source ↗Checked 2026-08-20 |
| H200CoreWeave8x H200 bare-metal GPU node | North America / Europe | on-demand | $6.31/GPU-hr50.44 | Provider source ↗Checked 2026-08-23 |
| B200CoreWeave8x B200 bare-metal GPU node | North America / Europe | on-demand | $8.60/GPU-hr68.80 | Provider source ↗Checked 2026-08-23 |
| H200Oracle Cloud8x H200 (BM.GPU.H200.8) | Select OCI regions | no commitment | $10.00/GPU-hr$10.00/GPU/hr (pay as you go) | Provider source ↗Checked 2026-08-20 |
| B200Oracle Cloud8x B200 (BM.GPU.B200.8) | Select OCI regions | no commitment | $14.00/GPU-hr$14.00/GPU/hr (pay as you go) | Provider source ↗Checked 2026-08-20 |
| B200Tscalebm-b200-8x-blackwell | Lagos, Nigeria | 1 month | Not publishedreserved capacity; quote required | Provider source ↗Checked 2026-08-23 |
| B200Impossible Cloud8x B200 bare-metal node | European Union | sales quote | Not publishedsales quote | Provider source ↗Checked 2026-08-23 |
| H200OpenMetal1x H200 NVL bare-metal server (starting configuration) | Ashburn, Virginia | monthly / sales quote | Not publishedmonthly / sales quote | Provider source ↗Checked 2026-09-08 |
| H200Impossible Cloud8x H200 bare-metal node | European Union | sales quote | Not publishedsales quote | Provider source ↗Checked 2026-08-23 |
| H200LeaderGPU1x H200 141GB bare-metal server | Netherlands | 1 day | Not published2986.9 | Provider source ↗Checked 2026-08-23 |
| H200Massed Compute8x H200 bare-metal cluster node | Not published | custom quote | Not publishedcustom quote | Provider source ↗Checked 2026-08-23 |
| H200DigitalOcean8x HGX H200 141GB SXM5 | NYC1 / AMS3 | sales contract | Not publishedcontract-based | Provider source ↗Checked 2026-09-08 |
| H200GlassPearl8x H200 whole bare-metal server | Not published | quote-only fixed whole-server contract | Not publishedquote-only fixed whole-server contract | Provider source ↗Checked 2026-08-23 |
| H200Sakura InternetKoukaryoku PHY: 8x H200 SXM 141GB | Ishikari, Japan | monthly / sales consultation | Not publishedmonthly / sales consultation | Provider source ↗Checked 2026-08-23 |
| B200Sakura InternetKoukaryoku PHY: 8x B200 SXM 180GB | Ishikari, Japan | monthly / sales consultation | Not publishedmonthly / sales consultation | Provider source ↗Checked 2026-08-23 |
| H200NAVER Cloud Platform8x H200 141GB bare-metal server | South Korea | quote | Not publishedquote | Provider source ↗Checked 2026-08-23 |
| B200GlassPearl8x HGX B200 whole bare-metal server | Not published | quote-only fixed whole-server contract | Not publishedquote-only fixed whole-server contract | Provider source ↗Checked 2026-08-23 |
Balance maturity and performance
H200 has a broader operational history. B200 can offer more performance and memory for supported workloads. Deployment readiness, software compatibility, delivery schedule, and provider experience belong in the decision.
Price the workload outcome
Benchmark representative prompts, sequences, batch sizes, precision, and communication patterns. Compare cost per trained token, generated token, or completed job alongside the infrastructure rate.
Is the premium worth paying?
The current NVIDIA B200 median is 92% above the NVIDIA H200 median in this marketplace dataset.
Pay the B200 premium when the workload can use Blackwell-specific throughput, its larger memory envelope, or its multi-GPU security properties enough to reduce completed-job cost or satisfy a hard requirement. Keep H200 as the baseline when software maturity, supplier choice, and operational familiarity dominate.
The median comparison is descriptive, not a performance benchmark. Configuration, region, and commitment mix can differ between the two offer sets.Questions buyers ask
Is B200 always cheaper per unit of work?
No. It depends on the workload, software maturity, achieved utilization, cluster design, and contract. Benchmark both configurations when the decision is material.