RTX 5090 workstation (1x 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.
NVIDIAcomponentAvailable17 models fit5 measured
Memory
32 GB
Official spec GDDR7
Memory bandwidth
1,792 GB/s
Official spec ~79% achieved in practice
Specifications
CPU
AMD Ryzen 9 9950X Official spec
GPU
GeForce RTX 5090 32 GB Official spec
Memory
32 GB GDDR7 Official spec
Memory bandwidth
1,792 GB/s Official spec source
Achieved bandwidth ⓘ
~79% (1,419 GB/s) Measured
FP16 compute ⓘ
~210 TFLOPS Official spec
System RAM
64 GB Official spec
Architecture
Discrete GPU with dedicated VRAM Official spec
Nominal power
700 W Official spec
Measured load power
640 W Measured
Idle power
70 W Measured
Released
30 Jan 2025 Official spec
Complete system price: card at roughly EUR 2,400 street plus about EUR 1,300 of platform. 1,792 GB/s makes anything that fits in 32 GB extremely fast; nothing above 32 GB fits without a large speed penalty. Achieved-bandwidth coefficient CALIBRATED to 79% from 3 measured runs on this configuration.
Standout models on this machine
Fastest
gpt-oss-20b — 419 t/s
MXFP4 · low confidence
Largest that fits
Compare against
Related hardware
Dual RTX 5090 workstation (2x 32 GB)64 GB · 3,584 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/sUsed RTX 4090 workstation (24 GB)24 GB · 1,008 GB/sRTX 3090 workstation (Ryzen 9 7950X, 64 GB)24 GB · 936 GB/sRadeon AI PRO R9700 workstation (32 GB)32 GB · 644 GB/s
Model performance on RTX 5090 workstation (1x 32 GB)5 measured, 12 estimated
17 of 17 rows
| Model↕ | Quant | Memory↕ | Decode▼ | Prefill↕ | Context | Fit | Confidence↕ |
|---|---|---|---|---|---|---|---|
| gpt-oss-20b OpenAI · 20.915B (3.6B active) | MXFP4recommended | 13.0 GB | 419 t/s | — | 128K | Comfortable | Low |
| Nemotron 3.5 Lightning 30B-A3B NVIDIA · 31.578B (3B active) | Q5_K_Mrecommended | 23.3 GB | ~319–459 t/s | ~3170–6580 t/s | 512K | Comfortable | Estimated |
| Qwen3-Coder 30B-A3B Alibaba Qwen · 30.532B (3.3B active) | Q5_K_Mrecommended | 23.2 GB | ~299–430 t/s | ~3070–6380 t/s | 64K | Comfortable | Estimated |
| Qwen3 30B-A3B Alibaba Qwen · 30.532B (3.3B active) | Q5_K_Mrecommended measured: Q4_K_M | 23.2 GB | ~299–430 t/s Measured 352.06 t/s | ~3070–6380 t/s | 40K evidence 0 | Comfortable | Estimated Measured · llama.cpp |
| Gemma 4 26B-A4B Google DeepMind · 25.806B (3.8B active) | Q5_K_Mrecommended | 21.0 GB | ~270–388 t/s | ~3110–6470 t/s | 32K | Comfortable | Estimated |
| Qwen3 8B Alibaba Qwen · 8.191B | Q8_0recommended | 10.6 GB | ~118–170 t/s | ~3760–7820 t/s | 40K | Comfortable | Estimated |
| Ternary Bonsai 2 27B Prism ML · 27.36B | PTQ1_0recommended measured: PQ2_0 | 7.2 GB | 121 t/s | 1810 t/s | 256K | Comfortable | Low |
| Phi-4 14B Microsoft · 14.66B | Q8_0recommended | 17.8 GB | ~69.6–100 t/s | ~2100–4370 t/s | 16K | Comfortable | Estimated |
| Gemma 3 27B Google DeepMind · 27.432B | Q4_K_Mrecommended | 21.3 GB | ~65.5–94.2 t/s | ~1120–2330 t/s | 16K | Comfortable | Estimated |
| Devstral Small 24B Mistral AI · 23.572B | Q5_K_Mrecommended | 18.8 GB | ~65–93.5 t/s | ~1310–2720 t/s | 64K | Comfortable | Estimated |
| Mistral Small 3.2 24B Mistral AI · 24.011B | Q5_K_Mrecommended | 19.1 GB | ~63.9–91.9 t/s | ~1280–2670 t/s | 64K | Comfortable | Estimated |
| Granite 4.2 30B IBM · 29.277B | Q4_K_Mrecommended | 20.5 GB | ~61.6–88.7 t/s | ~1050–2190 t/s | 32K | Comfortable | Estimated |
| Qwen3.8 27B Alibaba Qwen · 27.781B | Q5_K_Mrecommended | 21.0 GB | ~55.7–80.1 t/s | ~1110–2300 t/s | 128K | Comfortable | Estimated |
| Qwen3.6 27B Alibaba Qwen · 27.781B | Q5_K_Mrecommended measured: Q4_K_M | 21.0 GB | ~55.7–80.1 t/s Measured 71.26 t/s | ~1110–2300 t/s | 128K evidence 128 | Comfortable | Estimated Measured · llama.cpp |
| Qwen3 32B Alibaba Qwen · 32.762B | Q4_K_Mrecommended | 22.6 GB | ~55.4–79.7 t/s | ~941–1950 t/s | 16K | Comfortable | Estimated |
| DeepSeek-R1-Distill 32B DeepSeek · 32.764B | Q4_K_Mrecommended | 22.6 GB | ~55.4–79.7 t/s | ~941–1950 t/s | 16K | Comfortable | Estimated |
| Gemma 4 31B Google DeepMind · 31.273B | Q4_K_Mrecommended | 27.2 GB | 67.3 t/s | — | 8K | Borderline | High |
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 5 pairings on this machine that a real benchmark backs.
Submit a benchmarkContributions are reviewed before publication.
Ownership economicsUnited States (federal) · C corporation · 8h/day
Monthly economic cost
$46
Calculated after tax
Codex/Claude Code
$100/mo
≈ $100/mo · local is $54 less
Net cash at purchase
$3,332
Calculated VAT not reclaimable
Total over 5 years
$2,731
Calculated after tax, after resale
Cost per USD/1M tokens
$0.86
Calculated 53.1M 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.
Purchase-price input: $3,332 in United States (federal), tax/VAT excluded · estimated · NL retail (system build) (converted catalogue estimate). This is the localized purchase input; the after-tax economic cost is calculated separately below.
Monthly breakdown
| Depreciation 2,832 over 5 years, straight-line to a 500 residual | $47.21 |
| Electricity 32.4 kWh/month at 0.140/kWh | $4.53 |
| Cost of capital 4.0%/yr on 1,916 average capital employed | $6.39 |
| Monthly cost before tax | $58.13 |
| Electricity tax shield Running costs are deductible business expenses | −$0.95 |
| First-year expensing §179 (100.0%) | −$11.66 |
| Monthly economic cost after tax | $45.51 |
Three different numbers, deliberately
Cash cost
$3,332
Money that leaves the bank account on day one, net of reclaimable VAT.
Accounting depreciation
$47.21/month
$2,832 written down over 5 years to a $500 residual.
After-tax economic cost
$45.51/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
$500 (15%) Assumption
Two GPU generations later, the card is worth a fraction of its list price.
Electricity
$0.140/kWh Assumption
Average power draw
184 W Calculated
Load 640 W for 20% of powered hours, idle 70 W for the rest — a machine that is on is not generating tokens the whole time.
Investment allowances
§179 100.0% Official spec
Worth $700 in total — first-year expensing that replaces later tax depreciation.
Cost of capital
$6.39/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.