GPU Cloud
B300, dedicated.
A single NVIDIA B300 GPU, 288 GB HBM3e on the Blackwell Ultra architecture, on an x86 host. $6.49 per hour, billed per minute, the same terms as every other GPUwerk tier.
Rented first come, first served: deploy a B300 when capacity is free, or join its waitlist in the console.
Specifications
Per GPU. The GPU and its memory are yours alone for the rental.
| GPU | B300, Blackwell Ultra architecture |
|---|---|
| Memory | 288 GB HBM3e |
| Memory bandwidth | 8 TB/s |
| Precision | FP4 and FP8 with the second-generation Transformer Engine |
| Host | x86 host |
| Tenancy | The GPU is yours alone while you rent it, never shared |
GPU figures from NVIDIA HGX. GPUwerk publishes tok/s and training throughput figures only for runs on its own hardware. See the DGX Spark fleet benchmarks for how they are reported.
What would you run on it?
Blackwell Ultra with 288 GB per GPU: the most memory on one GPU in the GPUwerk catalog.
The largest single-GPU models
Up to about 288B parameters at FP8, or about 576B at FP4, by weights alone, on one GPU.
Reasoning and long context
Memory for the long KV caches that reasoning models and very long prompts build up.
Memory-bound training
Fine-tunes and training runs that run out of memory on a B200.
Poor fit: small or mid-size models
A model that fits in 80 to 141 GB costs less per hour on an H100 or H200.
Need all eight GPUs of one server? See the 8× B300 server.
What fits in 288 GB
Weights only: memory divided by bytes per parameter. Leave headroom for the KV cache, activations and the runtime, often 10 to 30% more.
| Precision | Per parameter | Largest model, weights only |
|---|---|---|
| FP16 / BF16 | 2 bytes | about 144B parameters |
| FP8 | 1 byte | about 288B parameters |
| FP4 (native on Blackwell Ultra) | 0.5 bytes | about 576B parameters |
Billing
| Rate | $6.49 per hour, billed per minute |
|---|---|
| Billing | Per minute, against a prepaid credit balance |
| Fees | No egress fees |
| Commitment | None |
Prices are in USD, excluding applicable tax. Full billing mechanics, including what a stop versus a terminate does, are on the pricing page.
FAQ
Can I rent a B300 now?
Deploy it from the console when capacity is free; while none is, the console offers a waitlist and people in line are emailed one at a time, oldest first. Capacity is not held for anyone: whoever deploys first gets it.
How does billing work on this tier?
The same as every GPUwerk tier: per minute, against a prepaid credit balance, in USD, excluding applicable tax, with no egress fees and no commitment.
Need GPU time today?
A DGX Spark has 128 GB of unified memory and is rentable now, at $0.79/hour billed per minute. First top-up: pay $20, get $30 in credit. Or deploy a B300.