DeepSeek-V4 Flash on NVIDIA DGX Station GB300 (748 GB)
304.18B (13B active) on 748 GB at 7,100 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
317.9 GB
Calculated of 635.8 GB usable — 50%
Max practical context
1M
Calculated model supports 1M
Memory budget at 8K context
Model weights ⓘ
307.0 GB Calculated
KV cache ⓘ
0.7 GB Calculated
Runtime overhead ⓘ
10.2 GB Estimated
Total required
317.9 GB Calculated
Usable memory ⓘ
635.8 GB Assumption
Headroom
317.9 GB Calculated
Highest-precision quantization that leaves headroom: uses 50% of usable memory at 8K context.
Unified memory: weights, KV cache and the OS share the same 748 GB pool. We assume 85% is addressable by the GPU, which matches a headless machine with the wired-memory limit raised.
Mixture of experts: all 304.18B parameters must be resident in memory even though only ~13B are active per token. Memory follows total parameters; speed follows active parameters.
PerformanceEstimated0/10
Decode (generation)
~50.4–72.6 t/s
Estimated calculated, not measured
Prefill (prompt)
~5690–11810 t/s
Estimated
TTFT at 8K
~0.8–1.6 s
Estimated time to first token
Power while generating
1350 W
Measured 0.15 tokens/s per watt
These figures are estimated, not measured. Nobody has published a benchmark of DeepSeek-V4 Flash 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
- Decode: reading 13.0B active parameters at 8.50 bits/weight takes 12.97 ms at 7100 GB/s x 15% achieved efficiency.
- MoE routing penalty of 15% applied: expert gathers are less bandwidth-efficient than a dense sweep.
- Prefill: 2500 TFLOPS (FP16) x 2 for native FP8 tensor cores x 22% assumed model-FLOPs utilisation, divided by 2 x 62.9B parameters per token.
- Prefill uses 62.9B effective parameters, not the 13B active in decode: a batch of hundreds of tokens routes across most of the expert pool.
- MoE decode depends on how well the engine batches expert gathers; real results vary more than for dense models.
- 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 | 165.7 GB | 172.4 GB | 27% | Comfortable | 1M |
Q4_K_M | 176.9 GB | 183.9 GB | 29% | Comfortable | 1M |
Q8_0recommended | 307.0 GB | 317.9 GB | 50% | Comfortable | 1M |
Cost of running DeepSeek-V4 Flash on this machineUnited States (federal) · C corporation · 8h/day
Monthly economic cost
$1,182
Calculated after tax
Codex/Claude Code
$100/mo
≈ $100/mo · local is $1,082 more
Net cash at purchase
$90,082
Calculated VAT not reclaimable
Total over 5 years
$70,927
Calculated after tax, after resale
Cost per USD/1M tokens
$151.67
Calculated 7.8M 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 79,272 over 5 years, straight-line to a 10,810 residual | $1,321.20 |
| Electricity 72.9 kWh/month at 0.140/kWh | $10.20 |
| Cost of capital 4.0%/yr on 50,446 average capital employed | $168.15 |
| Monthly cost before tax | $1,499.55 |
| Electricity tax shield Running costs are deductible business expenses | −$2.14 |
| First-year expensing §179 (100.0%) | −$315.29 |
| Monthly economic cost after tax | $1,182.12 |
Three different numbers, deliberately
Cash cost
$90,082
Money that leaves the bank account on day one, net of reclaimable VAT.
Accounting depreciation
$1,321.20/month
$79,272 written down over 5 years to a $10,810 residual.
After-tax economic cost
$1,182.12/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
$10,810 (12%) Assumption
Extrapolated.
Electricity
$0.140/kWh Assumption
Average power draw
414 W Calculated
Load 1350 W for 20% of powered hours, idle 180 W for the rest — a machine that is on is not generating tokens the whole time.
Investment allowances
§179 100.0% Official spec
Worth $18,917 in total — first-year expensing that replaces later tax depreciation.
Cost of capital
$168.15/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
0.7 t/s
Calculated purchase price only
Decode per $100/month
5.2 t/s
Calculated after-tax ownership cost
Tokens per joule
0.15
Calculated same as tokens/s per watt
USD per 1M output tokens
$151.67
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 | 767.6M/mo | $12 | API cheaper | Different model |
| DeepSeek-V4.1 Flash DeepSeek | $0.60 | 656.7M/mo | $14 | API cheaper | Different model |
| DeepSeek-V4 Pro DeepSeek | $1.98 | 162.8M/mo | $57 | API cheaper | Different model |
| DeepSeek V4 Pro (Together) Together AI | $4.40 | 55.8M/mo | $165 | API cheaper | Different model |
| GLM-5.3 Z.ai | $4.40 | 75.8M/mo | $122 | API cheaper | Different model |
| Claude Haiku 4.5 Anthropic | $5.00 | 90.9M/mo | $101 | API 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 $1,182/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-V4 Flash on other hardware
| Hardware | Decode | Price | Fit |
|---|---|---|---|
| NVIDIA DGX H200 (8x H200, 1,128 GB) | ~411–592 t/s | $359,430 * | Comfortable |
| Lenovo ThinkStation PX (4x RTX PRO 6000, 384 GB) | ~169–243 t/s | $58,553 * | Comfortable |
| Mac Studio M5 Ultra 256 GB | ~89.3–129 t/s | $8,647 * | Comfortable |
| Mac Studio M3 Ultra 256 GB | ~51.8–74.5 t/s | $5,674 * | Comfortable |
| Mac Studio M5 Ultra 512 GB | ~49.8–71.7 t/s | $11,350 * | Comfortable |
| Mac Studio M3 Ultra 512 GB | ~28.3–40.7 t/s | $8,287 * | Comfortable |
Nearest alternatives to the NVIDIA DGX Station GB300 (748 GB)
DeepSeek-V4 Flash on Mac Studio M5 Ultra 512 GB512 GBDeepSeek-V4 Flash on Mac Studio M3 Ultra 512 GB512 GBDeepSeek-V4 Flash on NVIDIA DGX H200 (8x H200, 1,128 GB)1,128 GBDeepSeek-V4 Flash on Mac Studio M5 Ultra 256 GB256 GBDeepSeek-V4 Flash on Mac Studio M3 Ultra 256 GB256 GB
Compare NVIDIA DGX Station GB300 (748 GB) against Mac Studio M5 Ultra 512 GB →
Compare NVIDIA DGX Station GB300 (748 GB) against Mac Studio M5 Ultra 512 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 (17 internal links) (+1)