Qwen3 8B on Dual RTX 5090 workstation (2x 32 GB)
8.191B on 64 GB at 3,584 GB/s. Hardware details · Model details
CompatibilityComfortable
Fits
Yes
Calculated Uses under 80% of usable memory. Room for a long context and other work.
Recommended quantization
Q8_0
Calculated gguf
Memory required
11.0 GB
Calculated of 56.3 GB usable — 20%
Max practical context
40K
Calculated model supports 40K
Memory budget at 8K context
Model weights ⓘ
8.3 GB Calculated
KV cache ⓘ
1.1 GB Calculated
Runtime overhead ⓘ
1.6 GB Estimated
Total required
11.0 GB Calculated
Usable memory ⓘ
56.3 GB Assumption
Headroom
45.3 GB Calculated
Highest-precision quantization that leaves headroom: uses 20% of usable memory at 8K context.
2 GPUs pooling 64 GB total VRAM. Assumes tensor- or layer-parallel sharding; each GPU carries its own context and communication buffers.
Multi-GPU throughput depends heavily on interconnect. Without NVLink, tensor parallelism over PCIe adds latency that partially offsets the extra bandwidth.
PerformanceEstimated0/10
Decode (generation)
~142–204 t/s
Estimated calculated, not measured
Prefill (prompt)
~6120–12700 t/s
Estimated
TTFT at 8K
~0.7–1.5 s
Estimated time to first token
Power while generating
1180 W
Measured 0.53 tokens/s per watt
These figures are estimated, not measured. Nobody has published a benchmark of Qwen3 8B on this machine that we know of, so we calculate throughput from memory bandwidth and active parameter count. Estimates are always shown as a range. Expand the working below to see exactly how the number was produced.
How the estimate is calculated
- Multi-GPU: 1792 GB/s per card, with each additional card contributing 40% of its bandwidth — 2509 GB/s effective, not the 3584 GB/s aggregate. Consumer Blackwell has no NVLink, so every layer's all-reduce crosses PCIe.
- Decode: reading 8.2B active parameters at 8.50 bits/weight takes 4.38 ms at 2509 GB/s x 79% achieved efficiency.
- Per-token overhead of 1.4 ms (kernel launches, attention bookkeeping, sampling) is significant here — this model is not purely bandwidth-bound on this hardware.
- Prefill: 419 TFLOPS (FP16) x 2 for native FP8 tensor cores x 0.71 calibrated against measured prefill on this platform x 26% assumed model-FLOPs utilisation, divided by 2 x 8.2B parameters per token.
- With few active parameters on fast memory, fixed per-token overhead rather than bandwidth sets the ceiling. Real engines vary widely in how well they hide it.
- Prefill throughput is highly engine-dependent. Flash attention, batch size and quantized KV all move this number substantially, and for sparse mixture-of-experts models it is the least reliable figure we produce.
- This is a calculated estimate, not a measurement. It assumes a single request, a short prompt, no speculative decoding and a warm model already resident in memory.
Estimator version estimator_v1. Stored with every estimated row so old estimates can be regenerated when the model improves.
This model at other quantizations on this machine
| Quantization | Weights | Total needed | Utilisation | Fit | Max context |
|---|---|---|---|---|---|
MLX 4-bit | 4.5 GB | 7.1 GB | 13% | Comfortable | 40K |
Q4_K_M | 4.8 GB | 7.4 GB | 13% | Comfortable | 40K |
Q8_0recommended | 8.3 GB | 11.0 GB | 20% | Comfortable | 40K |
BF16 | 15.3 GB | 18.2 GB | 32% | Comfortable | 40K |
Cost of running Qwen3 8B on this machineUnited 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
$3.65
Calculated 21.9M 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.
Efficiency metrics
Decode per $1,000 spent
29.5 t/s
Calculated purchase price only
Decode per $100/month
216.3 t/s
Calculated after-tax ownership cost
Tokens per joule
0.53
Calculated same as tokens/s per watt
USD per 1M output tokens
$3.65
Calculated at 20% utilisation
Local versus hosted APIs
| Hosted model | USD/1M output | Break-even | API at your volume | Verdict | Comparison type |
|---|---|---|---|---|---|
| Qwen3.8 Flash Alibaba Cloud | $0.42 | 51.9M/mo | $34 | API cheaper | Different model |
| DeepSeek-V4 Flash DeepSeek | $0.66 | 33.0M/mo | $53 | API cheaper | Different model |
| DeepSeek-V4 Pro DeepSeek | $1.98 | 11.0M/mo | $159 | Local cheaper | Different model |
| DeepSeek V4 Pro (Together) Together AI | $4.40 | 3.8M/mo | $465 | Local cheaper | Different model |
| GLM-5.3 Z.ai | $4.40 | 5.1M/mo | $342 | Local cheaper | Different model |
| Claude Haiku 4.5 Anthropic | $5.00 | 6.2M/mo | $285 | Local cheaper | Different model |
Different-model comparison: the hosted model is not the model you would run locally. Treat this as a workload-quality trade-off, not a direct economic equivalence — the frontier model may complete a task in fewer tokens, or complete tasks the local model cannot. Break-even is the monthly output volume at which API spend equals the $80/month economic cost of owning this machine, assuming 8 input tokens per output token. Hosted prices are published in USD. This table is the canonical US default.
Qwen3 8B on other hardware
| Hardware | Decode | Price | Fit |
|---|---|---|---|
| Quad RTX 5090 workstation (4x 32 GB) | ~196–282 t/s | €13.999 | Comfortable |
| RTX 5090 workstation (1x 32 GB) | ~118–170 t/s | €3.699 | Comfortable |
| RTX PRO 6000 Blackwell workstation (96 GB) | ~112–161 t/s | €11.499 | Comfortable |
| RTX PRO 6000 Max-Q workstation (96 GB, 300 W) | ~112–161 t/s | €11.999 | Comfortable |
| Mac Studio M5 Ultra 512 GB | ~88.4–127 t/s | €12.599 | Comfortable |
| Mac Studio M5 Ultra 256 GB | ~88.4–127 t/s | €9.599 | Comfortable |
Nearest alternatives to the Dual RTX 5090 workstation (2x 32 GB)
Qwen3 8B on RTX 5090 workstation (1x 32 GB)32 GBQwen3 8B on Quad RTX 5090 workstation (4x 32 GB)128 GBQwen3 8B on Dual used RTX 4090 workstation (48 GB)48 GBQwen3 8B on RTX PRO 6000 Blackwell workstation (96 GB)96 GBQwen3 8B on RTX PRO 6000 Max-Q workstation (96 GB, 300 W)96 GB
Compare Dual RTX 5090 workstation (2x 32 GB) against RTX 5090 workstation (1x 32 GB) →
Compare Dual RTX 5090 workstation (2x 32 GB) against RTX 5090 workstation (1x 32 GB) →
Page quality score (why this page is or is not indexed)
10/13 — indexed. Generated pages are gated so we do not ask a search engine to rank a page with nothing computed to say. The directive is emitted in the page head via the metadata API, not in the body, so it is authoritative.
- Has a memory-fit calculation (+3)
- Has a performance estimate (+2)
- Has a price (+2)
- Has an ownership economics calculation (+2)
- Well connected (18 internal links) (+1)