modelrigs

DeepSeek-R1-Distill 32B on NVIDIA DGX Spark 128 GB

32.764B on 128 GB at 273 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
37.1 GB
Calculated of 105.0 GB usable — 35%
Max practical context
128K
Calculated model supports 128K

Memory budget at 8K context

Model weights ⓘ
33.1 GB Calculated
KV cache ⓘ
2.0 GB Calculated
Runtime overhead ⓘ
2.0 GB Estimated
Total required
37.1 GB Calculated
Usable memory ⓘ
105.0 GB Assumption
Headroom
67.9 GB Calculated
Highest-precision quantization that leaves headroom: uses 35% of usable memory at 8K context.
Unified LPDDR: the accelerator allocates from the same 128 GB pool as the OS. We assume 82% is available to the inference engine.
PerformanceEstimated0/10
Decode (generation)
~4.3–6.2 t/s
Estimated calculated, not measured
Prefill (prompt)
~381–792 t/s
Estimated
TTFT at 8K
~10.4–21.6 s
Estimated time to first token
Power while generating
170 W
Measured 0.10 tokens/s per watt
These figures are estimated, not measured. Nobody has published a benchmark of DeepSeek-R1-Distill 32B 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
  1. Decode: reading 32.8B active parameters at 8.50 bits/weight takes 190.29 ms at 273 GB/s x 67% achieved efficiency.
  2. Prefill: 125 TFLOPS (FP16) x 2 for native FP8 tensor cores x 0.85 calibrated against measured prefill on this platform x 18% assumed model-FLOPs utilisation, divided by 2 x 32.8B parameters per token.
  • 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
QuantizationWeightsTotal neededUtilisationFitMax context
MLX 4-bit17.9 GB21.4 GB20%Comfortable128K
Q4_K_M19.1 GB22.6 GB22%Comfortable128K
Q8_0recommended33.1 GB37.1 GB35%Comfortable128K
BF1661.0 GB65.9 GB63%Comfortable96K
Cost of running DeepSeek-R1-Distill 32B on this machineUnited States (federal) · C corporation · 8h/day
Monthly economic cost
$52
Calculated after tax
Codex/Claude Code
$100/mo
≈ $100/mo · local is $48 less
Net cash at purchase
$3,873
Calculated VAT not reclaimable
Total over 5 years
$3,091
Calculated after tax, after resale
Cost per USD/1M tokens
$77.98
Calculated 0.7M 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
3,408 over 5 years, straight-line to a 465 residual
$56.80
Electricity
9.5 kWh/month at 0.140/kWh
$1.33
Cost of capital
4.0%/yr on 2,169 average capital employed
$7.23
Monthly cost before tax$65.36
Electricity tax shield
Running costs are deductible business expenses
−$0.28
First-year expensing
§179 (100.0%)
−$13.55
Monthly economic cost after tax$51.52

Three different numbers, deliberately

Cash cost
$3,873
Money that leaves the bank account on day one, net of reclaimable VAT.
Accounting depreciation
$56.80/month
$3,408 written down over 5 years to a $465 residual.
After-tax economic cost
$51.52/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
$465 (12%) Assumption
Extrapolated.
Electricity
$0.140/kWh Assumption
Average power draw
54 W Calculated
Load 170 W for 20% of powered hours, idle 25 W for the rest — a machine that is on is not generating tokens the whole time.
Investment allowances
§179 100.0% Official spec
Worth $813 in total — first-year expensing that replaces later tax depreciation.
Cost of capital
$7.23/month Assumption
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
1.3 t/s
Calculated purchase price only
Decode per $100/month
10.1 t/s
Calculated after-tax ownership cost
Tokens per joule
0.10
Calculated same as tokens/s per watt
USD per 1M output tokens
$77.98
Calculated at 20% utilisation
Local versus hosted APIs
Hosted modelUSD/1M outputBreak-evenAPI at your volumeVerdictComparison type
Qwen3.8 Flash
Alibaba Cloud
$0.4233.5M/mo$1API cheaperDifferent model
DeepSeek-V4 Flash
DeepSeek
$0.6621.3M/mo$2API cheaperDifferent model
DeepSeek-V4 Pro
DeepSeek
$1.987.1M/mo$5API cheaperDifferent model
DeepSeek V4 Pro (Together)
Together AI
$4.402.4M/mo$14API cheaperDifferent model
GLM-5.3
Z.ai
$4.403.3M/mo$10API cheaperDifferent model
Claude Haiku 4.5
Anthropic
$5.004.0M/mo$9API cheaperDifferent 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 $52/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.
DeepSeek-R1-Distill 32B on other hardware
HardwareDecodePriceFit
Quad RTX 5090 workstation (4x 32 GB)~65.3–94 t/s€13.999Comfortable
RTX 5090 workstation (1x 32 GB)~55.4–79.7 t/s€3.699Comfortable
Dual RTX 5090 workstation (2x 32 GB)~43.3–62.3 t/s€6.499Comfortable
Dual used RTX 4090 workstation (48 GB)~42.2–60.7 t/s€4.099Comfortable
Used RTX 4090 workstation (24 GB)~34.7–49.9 t/s€2.299Borderline
RTX PRO 6000 Blackwell workstation (96 GB)~30.4–43.8 t/s€11.499Comfortable
Nearest alternatives to the NVIDIA DGX Spark 128 GB
DeepSeek-R1-Distill 32B on Framework Desktop (Ryzen AI Max+ 395, 128 GB)128 GBDeepSeek-R1-Distill 32B on GMKtec EVO-X2 (Ryzen AI Max+ 395, 128 GB)128 GBDeepSeek-R1-Distill 32B on Quad RTX 5090 workstation (4x 32 GB)128 GBDeepSeek-R1-Distill 32B on Mac Studio M5 Max 128 GB128 GBDeepSeek-R1-Distill 32B on Dual used RTX 4090 workstation (48 GB)48 GB
Compare NVIDIA DGX Spark 128 GB against Framework Desktop (Ryzen AI Max+ 395, 128 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)