Dual RTX 5090 workstation (2x 32 GB)
The fastest consumer memory subsystem you can buy: 1,792 GB/s of GDDR7. Only 32 GB of it, which is the whole story of this card for local AI — extremely fast on anything that fits, useless on anything that does not.
NVIDIAcomponent15 models fit0 measured
Memory
64 GB
Official spec GDDR7
Memory bandwidth
3,584 GB/s
Official spec ~79% achieved in practice
Specifications
CPU
AMD Ryzen 9 9950X Official spec
GPU
2× 2x GeForce RTX 5090 32 GB Official spec
Memory
64 GB GDDR7 Official spec
Memory bandwidth
3,584 GB/s Official spec source
Achieved bandwidth ⓘ
~79% (2,838 GB/s) Assumption
FP16 compute ⓘ
~419 TFLOPS Official spec
System RAM
128 GB Official spec
Architecture
Discrete GPU with dedicated VRAM Official spec
Nominal power
1400 W Official spec
Measured load power
1180 W Measured
Idle power
110 W Measured
Released
30 Jan 2025 Official spec
64 GB of very fast VRAM, but no NVLink on GeForce Blackwell: tensor parallelism runs over PCIe 5.0 and single-stream decode does not reach the aggregate 3.6 TB/s. Needs a 1,600 W PSU and produces roughly 1.2 kW of heat under load. Achieved-bandwidth coefficient INHERITED from rtx-5090-workstation, which shares the same silicon and software stack.
Standout models on this machine
Fastest
gpt-oss-20b — ~324–466 t/s
MXFP4 · estimated confidence
Largest that fits
Compare against
Related hardware
RTX 5090 workstation (1x 32 GB)32 GB · 1,792 GB/sQuad RTX 5090 workstation (4x 32 GB)128 GB · 7,168 GB/sDual used RTX 4090 workstation (48 GB)48 GB · 2,016 GB/sRTX PRO 6000 Blackwell workstation (96 GB)96 GB · 1,792 GB/sRTX PRO 6000 Max-Q workstation (96 GB, 300 W)96 GB · 1,792 GB/sUsed RTX 4090 workstation (24 GB)24 GB · 1,008 GB/s
Model performance on Dual RTX 5090 workstation (2x 32 GB)0 measured, 15 estimated
15 of 15 rows
| Model↕ | Quant | Memory↕ | Decode▼ | Prefill↕ | Context | Fit | Confidence↕ |
|---|---|---|---|---|---|---|---|
| gpt-oss-20b OpenAI · 20.915B (3.6B active) | MXFP4 | 13.4 GB | ~324–466 t/s | ~11550–23980 t/s | 128K | Comfortable | Estimated |
| Qwen3-Coder 30B-A3B Alibaba Qwen · 30.532B (3.3B active) | Q8_0 | 33.9 GB | ~236–339 t/s | ~4990–10370 t/s | 128K | Comfortable | Estimated |
| Qwen3 30B-A3B Alibaba Qwen · 30.532B (3.3B active) | Q8_0 | 33.9 GB | ~236–339 t/s | ~4990–10370 t/s | 40K | Comfortable | Estimated |
| Gemma 4 26B-A4B Google DeepMind · 25.806B (3.8B active) | Q8_0 | 30.1 GB | ~216–311 t/s | ~5060–10510 t/s | 64K | Comfortable | Estimated |
| Qwen3 8B Alibaba Qwen · 8.191B | Q8_0 | 11.0 GB | ~142–204 t/s | ~6120–12700 t/s | 40K | Comfortable | Estimated |
| Phi-4 14B Microsoft · 14.66B | Q8_0 | 18.2 GB | ~88.7–128 t/s | ~3420–7100 t/s | 16K | Comfortable | Estimated |
| Devstral Small 24B Mistral AI · 23.572B | Q8_0 | 27.2 GB | ~58.5–84.2 t/s | ~2130–4410 t/s | 128K | Comfortable | Estimated |
| Mistral Small 3.2 24B Mistral AI · 24.011B | Q8_0 | 27.6 GB | ~57.6–82.8 t/s | ~2090–4330 t/s | 128K | Comfortable | Estimated |
| Gemma 3 27B Google DeepMind · 27.432B | Q8_0 | 33.8 GB | ~51–73.4 t/s | ~1830–3790 t/s | 32K | Comfortable | Estimated |
| Qwen3.8 27B Alibaba Qwen · 27.781B | Q8_0 | 32.3 GB | ~50.4–72.6 t/s | ~1800–3750 t/s | 64K | Comfortable | Estimated |
| Qwen3.6 27B Alibaba Qwen · 27.781B | Q8_0 | 32.3 GB | ~50.4–72.6 t/s | ~1800–3750 t/s | 64K | Comfortable | Estimated |
| Gemma 4 31B Google DeepMind · 31.273B | Q8_0 | 41.4 GB | ~45.2–65.1 t/s | ~1600–3330 t/s | 16K | Comfortable | Estimated |
| Qwen3 32B Alibaba Qwen · 32.762B | Q8_0 | 37.5 GB | ~43.3–62.4 t/s | ~1530–3180 t/s | 32K | Comfortable | Estimated |
| DeepSeek-R1-Distill 32B DeepSeek · 32.764B | Q8_0 | 37.5 GB | ~43.3–62.3 t/s | ~1530–3180 t/s | 32K | Comfortable | Estimated |
| Llama 3.3 70B Meta · 70.554B | MLX 4-bit | 43.5 GB | ~38.4–55.2 t/s | ~1420–2950 t/s | 16K | Comfortable | Estimated |
Rows in grey are estimates from our bandwidth model, not measurements — they are shown as a range and never as a precise figure. Use the “measured only” filter to see just the 0 pairings on this machine that a real benchmark backs.
Ownership economicsUnited States (federal) · C corporation · 8h/day
Monthly economic cost
$80
Calculated after tax
Codex/Claude Code
$100/mo
≈ $100/mo · local is $20 less
Net cash at purchase
$5,854
Calculated VAT not reclaimable
Total over 5 years
$4,799
Calculated after tax, after resale
Cost per USD/1M tokens
$1.60
Calculated 50.0M tokens/month
The $100 comparison uses the Codex Pro 5x / Claude Max 5x plans. A fixed planning conversion is used for USD. This compares monthly spend only: subscriptions have usage limits, local hardware has different capabilities and constraints, and taxes or regional pricing may change the charged amount. Prices checked 7 September 2026.
Monthly breakdown
| Depreciation 4,976 over 5 years, straight-line to a 878 residual | $82.94 |
| Electricity 57.0 kWh/month at 0.140/kWh | $7.98 |
| Cost of capital 4.0%/yr on 3,366 average capital employed | $11.22 |
| Monthly cost before tax | $102.14 |
| Electricity tax shield Running costs are deductible business expenses | −$1.68 |
| First-year expensing §179 (100.0%) | −$20.49 |
| Monthly economic cost after tax | $79.98 |
Three different numbers, deliberately
Cash cost
$5,854
Money that leaves the bank account on day one, net of reclaimable VAT.
Accounting depreciation
$82.94/month
$4,976 written down over 5 years to a $878 residual.
After-tax economic cost
$79.98/month
Depreciation plus running costs plus cost of capital, less the tax those deductions save. This is the figure to compare between machines.
Assumptions and sources (verified 2026-09-07)
VAT rate
0% Official spec
VAT recoverable
0% Official spec
Federal C-corporation estimate at the flat 21% rate. Pass-through entities should use the sole-proprietor estimate as a rougher proxy.
Effective deduction rate
21.00% Calculated
Headline marginal rate 21.00%.
Depreciation
5 years, straight-line Assumption
Residual value
$878 (15%) Assumption
Two GPU generations later, the card is worth a fraction of its list price.
Electricity
$0.140/kWh Assumption
Average power draw
324 W Calculated
Load 1180 W for 20% of powered hours, idle 110 W for the rest — a machine that is on is not generating tokens the whole time.
Investment allowances
§179 100.0% Official spec
Worth $1,229 in total — first-year expensing that replaces later tax depreciation.
Cost of capital
$11.22/month Assumption
- VAT rate: No US federal VAT · verified 2026-09-07
- Marginal tax rate: IRS Publication 542 — Corporations · verified 2026-09-07
- VAT recoverable fraction: site assumption · verified 2026-09-07
- Useful life: IRS Publication 946 — How To Depreciate Property · verified 2026-09-07
- Electricity price: Site assumption · verified 2026-09-07
Assumes the machine generates tokens 20% of its 176 powered hours per month. Cost per token scales inversely with this number — halve the utilisation and the cost per token doubles.
Federal planning estimate only, not tax advice. State and local income tax, sales/use tax and incentives are excluded. It assumes 100% business use, a Section 179 election, enough business income to use it, and that the full annual limit remains available.