GPU price check · updated 2026-09-10
B200 rental prices
B200 rents for $6.00–$14.24 per GPU-hour on demand across 8 sellers this week. The spread is not noise — it is mostly the network fabric, which determines what the chip can actually do.
$6.00
lowest on-demand
$7.67
median
4
Grade A/B offerings
0
unverifiable listings
Fair-price band (verifiable training-class, Grade A/B): $7.15–$8.39 per GPU-hour on demand. Below that band, ask what the seller is not telling you about the fabric; above it, you are paying for something the rate card should be able to name.
Every listing we track
| Provider | SKU | $/GPU-hr | Terms | Grade | Fabric (as published) | Src |
|---|---|---|---|---|---|---|
| Nebius | NVIDIA HGX B200 — preemptible | $3.95 | spot | B | InfiniBand, 400 Gbps per GPU / 3.2 Tbps per node (per docs.nebius.com) | ↗ |
| Hyperstack | NVIDIA B200 SXM | $6.00 | on-demand | C | not published | ↗ |
| DataCrunch (rebranded to Verda, verda.com) | B200 SXM6 | $6.11 | on-demand | C | not published | ↗ |
| Lambda | 8x NVIDIA B200 SXM6 (on-demand instance) | $6.69 | on-demand | C | not published | ↗ |
| Together AI (GPU Clusters) | HGX B200 (reserved) | $6.79 | reserved | B | High-speed InfiniBand | ↗ |
| Nebius | NVIDIA HGX B200 (gpu-b200-sxm platform) | $7.15 | on-demand | A | InfiniBand, 400 Gbps per GPU / 3.2 Tbps per 8-GPU node, GPUDirect RDMA (per docs.nebius.com) | ↗ |
| Together AI (GPU Clusters) | HGX B200 | $8.19 | on-demand | B | High-speed InfiniBand | ↗ |
| CoreWeave | NVIDIA HGX B200 (8x) | $8.60 | on-demand | B | 8x NVIDIA ConnectX-7 InfiniBand HCAs, 400G NDR non-blocking NVIDIA Quantum-2 fabric, plus BlueField-3 DPU (per B200 release notes) | ↗ |
| Lambda | 1-Click Cluster — NVIDIA HGX B200 | $9.86 | reserved | B | NVIDIA Quantum-2 InfiniBand with SHARP acceleration (per-node bandwidth not stated on fetched pages) | ↗ |
| Oracle Cloud (OCI) | BM.GPU.B200.8 | $14.00 | on-demand (PAY_AS_YOU_GO) | C | RDMA cluster network: '8x400 Gb/sec' = 3,200 Gbps per node (Oracle price-list shape table; fabric protocol not named for this row) | ↗ |
| AWS | p6-b200.48xlarge | $14.24 | on-demand | A | EFAv4, 3.2 Tbps per instance (AWS P6 page) | ↗ |
| Google Cloud | a4-highgpu-8g | — | — | C | RoCE via NVIDIA ConnectX-7 NICs on Google's four-way rail-aligned network, 3,600 Gbps max per VM, 10 NICs (Google doc) | ↗ |
Grades summarize deliverability from the seller's own published specs — methodology. A missing fabric field is recorded as unknown, never guessed. Weekly context in the Delivered Compute Report.
Prices move weekly. Get the report.
Headline prices, deliverability grades, and one teardown a week — free.
Subscribe free