NVIDIA H100
Verified offers33
Price floor$1.68per GPU-hour
Median$2.71per GPU-hour
Choose H100 when the workload fits its memory envelope and price, region, or near-term availability is the binding constraint.
Review NVIDIA H100 pricesNVIDIA H100 vs H200 prices
H100 offers broad current supply. H200 adds memory capacity and bandwidth. The right choice depends on delivered workload cost and availability.
Bottom line
H100 is the safer default when price and current supplier choice matter most. H200 earns its premium when 141 GB of memory or higher memory bandwidth materially reduces partitioning or increases delivered throughput.
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 H200
Verified offers20
Price floor$2.51per GPU-hour
Median$3.12per GPU-hour
Choose H200 when additional memory or bandwidth removes a real bottleneck and improves completed-workload economics.
Review NVIDIA H200 prices| Decision factor | NVIDIA H100 | NVIDIA H200 |
|---|---|---|
| Architecture | Hopper | Hopper |
| GPU memory | 80 GB HBM3 (SXM); 94 GB HBM3 on H100 NVL | 141 GB HBM3e |
| Memory bandwidth | 3.35 TB/s (SXM); 3.9 TB/s on H100 NVL | 4.8 TB/s |
| Typical topology | HGX partner systems with 4 or 8 SXM GPUs; NVL systems vary by provider | HGX H200 partner systems with 4 or 8 GPUs; H200 NVL options also exist |
| Marketplace availability | Broader verified supply in the current marketplace index | Fewer verified offers than H100 in the current marketplace index |
| Workload fit | Training and inference that fit within 80–94 GB per GPU, with broad supplier choice | Memory-bound inference, longer contexts, larger batches, and models that otherwise need more partitioning |
Reference specifications: NVIDIA H100 specifications ↗ · NVIDIA H200 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 |
|---|---|---|---|---|
| H100Latitude.sh1x H100 80GB (g3.h100.small) | Not published | no commitment | $1.68/GPU-hr$1.68/hr ($1,230/mo) | Provider source ↗Checked 2026-08-20 |
| H100Qubrid AI8x H100 80GB bare-metal server | Not published | 12 months | $1.90/GPU-hr133414.80 | Provider source ↗Checked 2026-08-23 |
| H100Voltage Park8x HGX H100 SXM | Not published | no commitment | $1.99/GPU-hrFrom $1.99/GPU/hr (no contract) | Provider source ↗Checked 2026-08-20 |
| H100Leaseweb4x H100 (HP DL385 G11) | London | term not stated | $2.02/GPU-hr£4,337.86/mo | Provider source ↗Checked 2026-08-20 |
| H100Tscalebm-h100-8x-sxm5 | Lagos, Nigeria | 1 month | $2.05/GPU-hr1-year reserve effective rate | Provider source ↗Checked 2026-08-23 |
| H100Vultr8x H100 80GB SXM (HGX H100) | Not published | term not published | $2.30/GPU-hr$2.300/GPU/hr (pre-paid contract) | Provider source ↗Checked 2026-08-20 |
| H100FluidCore8x H100 80GB bare-metal node | Not published | hourly | $2.36/GPU-hr18.88 | Provider source ↗Checked 2026-08-23 |
| 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 |
| H100Taiga Cloud8x H100 SXM | UK / PT / SE / NO | hourly | $2.60/GPU-hrFrom $2.60/GPU/hr (bare-metal tier) | Provider source ↗Checked 2026-09-04 |
| H100Leaseweb1x H100 (HP DL385 G11) | Frankfurt | term not stated | $2.61/GPU-hr€1,638.57/mo | Provider source ↗Checked 2026-08-20 |
| H100Gcore8x H100 SXM (bm3-ai-1xlarge-h100-80-8) | Not published | hourly | $2.64/GPU-hr18.16 | Provider source ↗Checked 2026-09-04 |
| H100ScalewayEM-T620E-H100: 2x H100 80GB | PAR-2 | monthly | $2.78/GPU-hr3499.99 | Provider source ↗Checked 2026-09-08 |
| 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 |
| H100Hostrunway1x H100 (GPU Montreal 5) | Montreal | no lock-in | $3.25/GPU-hr$2,370/mo | Provider source ↗Checked 2026-08-20 |
| H100Hostrunway1x H100 dedicated GPU server (dual EPYC 7543) | Montreal | no lock-in; 15% discount at 12 months | $3.42/GPU-hr2496.00 | Provider source ↗Checked 2026-08-29 |
| H100Hostrunway3x H100 dedicated GPU server (dual EPYC 9334) | Montreal | no lock-in; 15% discount at 12 months | $3.75/GPU-hr8210.00 | Provider source ↗Checked 2026-08-29 |
| H100Cirrascale8x H100 SXM | US | 1-mo min | $4.28/GPU-hr$24,999/mo (annual commit $19,999/mo) | Provider source ↗Checked 2026-08-20 |
| H100Hostrunway1x H100 dedicated GPU server (dual EPYC 9224) | Montreal | no lock-in; 15% discount at 12 months | $4.36/GPU-hr3180.00 | Provider source ↗Checked 2026-08-29 |
| 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 |
| H100Yotta Data Services / Shakti CloudBM Large: 8x H100 80GB HGX bare-metal server | India | on-demand | $4.99/GPU-hr39.92 | Provider source ↗Checked 2026-08-23 |
| H100Hostrunway1x H100 dedicated GPU server (dual EPYC 9224) | Germany | no lock-in; 15% discount at 12 months | $5.55/GPU-hr4050.00 | Provider source ↗Checked 2026-08-29 |
| H100CoreWeave8x H100 bare-metal GPU node | North America / Europe | on-demand | $6.16/GPU-hr49.24 | Provider source ↗Checked 2026-08-23 |
| H200CoreWeave8x H200 bare-metal GPU node | North America / Europe | on-demand | $6.31/GPU-hr50.44 | 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 |
| H100Oracle Cloud8x H100 (BM.GPU.H100.8) | Select OCI regions | no commitment | $10.00/GPU-hr$10.00/GPU/hr (pay as you go) | Provider source ↗Checked 2026-08-20 |
| H100Denvr AI Cloud8x H100 80GB SXM bare-metal full node | Calgary MSC1 / Houston HOU1 | sales reservation | Not publishedsales reservation | 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 |
| H100ScalewayUp to 8x H100 | PAR-2 | no min stated | Not publishedNot published | Provider source ↗Checked 2026-08-20 |
| H100GPU Server Rentals8x HGX H100 SXM5 bare-metal node | Not published | month-to-month available; typical 12-36 months | Not publishedquote-only | Provider source ↗Checked 2026-08-23 |
| H100Impossible Cloud8x H100 bare-metal node | European Union | sales quote | Not publishedsales quote | Provider source ↗Checked 2026-08-23 |
| H100LeaderGPU8x H100 80GB NVLink bare-metal server | Netherlands | 1 day | Not published11900 | Provider source ↗Checked 2026-08-23 |
| H200LeaderGPU1x H200 141GB bare-metal server | Netherlands | 1 day | Not published2986.9 | Provider source ↗Checked 2026-08-23 |
Showing 40 of 53 matching records. Browse all listings.
Choose from workload constraints
Start with model memory, context length, batch size, precision, parallelism, and interconnect needs. H200 can reduce partitioning or support larger batches; H100 can provide wider supplier choice and attractive pricing.
Compare contracts with the hardware
Normalize the rate, then preserve provider, region, term, billing basis, node topology, network, and delivery date. A lower rate attached to an unsuitable commitment or topology does not reduce completed-workload cost.
Is the premium worth paying?
The current NVIDIA H200 median is 15% above the NVIDIA H100 median in this marketplace dataset.
Pay the H200 premium only when a representative benchmark shows that its larger, faster memory lowers completed-workload cost or enables a workload that does not fit the selected H100 configuration. Otherwise, H100 usually leaves more room to optimize provider, region, and contract terms.
The median comparison is descriptive, not a performance benchmark. Configuration, region, and commitment mix can differ between the two offer sets.Questions buyers ask
Should I rent H100 or H200?
Select the least expensive configuration that meets model memory, throughput, topology, availability, and security requirements after benchmarking the workload.