1 live offerings · 1 tracked regions
Available now 1
Loading
Loading workspaces…
engineering · GPU0107
Specifications with their source, supported precisions, interconnect, benchmark coverage and which published models fit NVIDIA B200 SXM 192GB — GMI Cloud identity.
| Specification | Value | Source | Kovara registry |
|---|---|---|---|
| Memory capacity | 192GB | Kovara registry · verified 2026-09-23 | used |
| Memory type | HBM3E | Kovara registry · verified 2026-09-23 | used |
| Memory bandwidth | 8TB/s | Kovara registry · verified 2026-09-23 | used |
| Board power (TDP) | 1,000W | Kovara registry · verified 2026-09-23 | used |
| Form factor | SXM | Kovara registry · verified 2026-09-23 | used |
| NVLink | Yes | Kovara registry · verified 2026-09-23 | used |
| MIG partitioning | Yes | Kovara registry · verified 2026-09-23 | used |
No public benchmark runs yet
Nobody has published a hardware run on this accelerator. Measured numbers appear here once a run of a compute repository is published.
AcceleratorsGPU0107
NVIDIA
Core hardware and lifecycle, drawn from Kovara master data.
What the recorded specifications mean in practice, and what this accelerator is used for.
Specs, explained
HBM (High Bandwidth Memory) is stacked in layers and sits right next to the processor on the same package. That short, wide path is what gives accelerators with HBM their memory bandwidth, so they are usually chosen for work limited by how fast data reaches the cores, such as large-model inference and HPC simulation.
VRAM is the working space on the card. Model weights, activations and datasets have to fit in it, or be split across several GPUs. Rule of thumb for the weights alone at 16-bit precision (about 2 GB per billion parameters): up to roughly 96 billion parameters, before activations and cache. This is arithmetic on the recorded VRAM, not a benchmark.
Blackwell added fifth-generation Tensor Cores and support for lower-precision formats down to FP4. (General context about the architecture, not a statement about this specific card.)
Launched in 2024. Kovara's records list it as a current-generation part.
NVLink is a direct GPU-to-GPU link. Where it is supported, GPUs in the same server can exchange data faster than over PCIe alone, which matters when one training job is spread across several GPUs.
Multi-Instance GPU (MIG) can divide one physical GPU into isolated slices, each with its own memory and compute, so several smaller workloads can share a card.
SXM modules are mounted directly on a server baseboard rather than slotted in as cards, the layout typically used for dense multi-GPU servers.
TDP is the rated thermal design power: the planning figure for power delivery and cooling in a server.
These are the uses Kovara has recorded for this part. The notes explain the workload in general terms; they are not performance claims.
Figures are Kovara master data, last verified 23 Sept 2026. Explanations describe the technology in general; they are not benchmarks and say nothing about price or availability.
Explore this accelerator on ComputePublic Kovara observations; not a claim of complete or universal market coverage. Incompatible raw price units are excluded.
Observed GPU-hour range
USD / GPU-hour
Spread $1.00 USD / GPU-hour
Providers represented by current Kovara-tracked offerings.
1 live offerings · 1 tracked regions
Available now 1
Regions represented by current offerings; this does not describe complete global availability.
1 live offerings · 1 providers
Current configurations, recorded availability, confirmed capacity and the evidence behind each price.
| Provider | Region | Configuration | Availability | Price | Commercial model | Trust / evidence | Action |
|---|---|---|---|---|---|---|---|
| OFR-0751GMI Cloud | Global / region-aware, Multiple / Unspecified | NVIDIA B200 SXM 192GB — GMI Cloud identity1 GPU configuration · 1 GPU per node | Available nowConfirmed capacity: Not confirmed | $5.00 USD / GPU-hourOriginal: USD 5.00 GPU/hour | On-DemandDedicated GPU · Provider-published from price · On-Demand commercial basis | Confidence AChecked 17 days agoConfirmed 29 Sept 2026Source: SRC000026 | Inspect |
Recorded hardware attributes. Not tracked is distinct from No.
Public evidence and confidence signals supporting Kovara-tracked coverage.