Skip to content

NVIDIATowerPrivate use

Workstation RTX 5090

Up to two 32 GB Blackwell cards in a Threadripper tower.

Starting at $12,506Built to order~1 week

Workstation RTX 5090

Plan your model.

Memory planning for reviewed model profiles.

See every model

  1. Qwen3 32B

    ollama · Precision not pinned32 GB threshold · 32 GB capacity64 GB system RAM required

  2. Qwen3 Coder 30B

    ollama · Precision not pinned32 GB threshold · 32 GB capacity64 GB system RAM required

Catalog thresholds are planning guidance, not measured peak memory. Confirm the artifact, precision, system RAM and runtime before deployment.

Build yours.

Workstation RTX 5090

1× 32 GBInstalled GPU memory · 32 GB

System RAM
64 GB
Delivery
Hosted in Oregon
Use
Private use

GPUs

1× 32 GB = 32 GB of VRAM

System memory

Storage

Delivery

Custom CPU, board, or more than the configurator allows?

Your details

What's inside.

B3IQ hand-builds it on a WRX90 Threadripper PRO board, with RDIMM up to 512 GB, and runs every card at full load before it leaves the bench.

GPU
1×NVIDIA RTX 5090 32GB
NVIDIA's Blackwell consumer flagship, for render, diffusion and inference.
Memory per GPU
32GB GDDR7
1.8 TB/s memory bandwidth per GPU. Check the runtime profile for your workload's memory requirements.
CPU
Threadripper PRO 9955WX (16c)
The selected build lists the exact processor and board.
Motherboard
WRX90 workstation board
A tower board, with the PSU and cooling to match.
Cooling
Tuned workstation tower airflow
High-airflow tower cooling.
Power
~1.8 kW (1×) to ~2.4 kW (2×)
One PSU, sized for the build.
System memory
64 GB DDR5 RDIMM
Storage
4 TB NVMe

Keep exploring.

All machines
  • RTX 5090

    TowerPrivate use

    RTX 5090

    32 GB Blackwell for your local AI and rendering workloads.

    Rent

    Rental quote

    By the day or 2–⁠24 months

    Own

    Starting at

    $7,880

    Built to order · ~1 week

    Ask about rentingConfigure
    • VRAM: 32 GB GDDR7
    • Reference configuration: 1 GPU
    • System RAM: 64 GB
    • Storage: 2 TB NVMe
    • Bandwidth: 1.8 TB/sMemory bandwidth
  • RTX PRO 6000 Server Edition

    Hosted rackCan earn

    RTX PRO 6000 Server Edition

    96 GB per GPU for memory-heavy inference. Check the runtime and cache budget.

    Rent

    Rental quote

    By the day or 2–⁠24 months

    Own

    Starting at

    $99,971

    Built to order · 6–⁠8 weeks

    Ask about rentingConfigure
    • VRAM: 96 GB GDDR7
    • Reference configuration: 4 GPUs
    • System RAM: 512 GB
    • Storage: 4 TB NVMe
    • Bandwidth: 1.6 TB/sMemory bandwidth

Deployment planning

Plan your model deployment.

These catalog profiles can be sized to this machine's GPU memory and system RAM. Check the required configuration in each guide; compatibility and performance still need testing.

Qwen3 32B

ollama · 32 GB catalog GPU-memory threshold · 1-GPU planning configuration

Qwen3 Coder 30B

ollama · 32 GB catalog GPU-memory threshold · 1-GPU planning configuration

gpt-oss 20B

ollama · 24 GB catalog GPU-memory threshold · 1-GPU planning configuration

Gemma 3 27B

ollama · 24 GB catalog GPU-memory threshold · 1-GPU planning configuration

Browse model guides

Reviewed model identities, runtime profiles and source evidence.

Find hardware by memory

Estimate the memory budget for the workload you want to run.

Compare providers and GPUs

Compare purchasing models, operating responsibilities and hardware.

How we review model evidence

Catalog settings, memory estimates and measured evidence have different limits.

Daryl XuViktoriya HyingSean GengMatt

Questions before you buy?

Invoicing, net terms, a custom build, a BAA or a security review. We'll sort it out on a call.

Book a callExplore machines

Or message us: iMessage +1 (332) 324-8327Telegram @b3iqbot