Devstral Small 24B on Lenovo ThinkStation PX (4x RTX PRO 6000, 384 GB)
23.572B on 384 GB at 7,168 GB/s. Hardware details · Model details
Yes. Devstral Small 24B fits on Lenovo ThinkStation PX (4x RTX PRO 6000, 384 GB) at the recommended
Q8_0 configuration, requiring approximately 28.0 GB at 8K context. Practical context capacity is 128K. Expected decode for the recommended configuration is ~76.3–110 t/s. EstimatedCompatibilityComfortable
Calculated fit
Yes
Calculated Uses under 80% of usable memory. Room for a long context and other work.
Recommended quantization
Q8_0
Calculated gguf
Memory required
28.0 GB
Calculated of 337.9 GB usable — 8%
Max practical context
128K
Calculated model supports 128K
Memory budget at 8K context
Model weights ⓘ
23.8 GB Calculated
KV cache ⓘ
1.3 GB Calculated
Runtime overhead ⓘ
2.9 GB Estimated
Total required
28.0 GB Calculated
Headroom ⓘ
310.0 GB Calculated
Highest-precision quantization that leaves headroom: uses 8% of usable memory at 8K context.
4 GPUs providing 384 GB aggregate VRAM. Assumes tensor- or layer-parallel sharding; each GPU carries its own context and communication buffers. Interconnect: PCIe 5.0 x16.
Multi-GPU throughput depends heavily on interconnect. Without NVLink or NVSwitch, tensor parallelism over PCIe adds latency that partially offsets the extra bandwidth.
Performance
Estimated performance · recommended Q8_0
Decode, prefill and TTFT below are estimates for Q8_0. We do not have a comparable Q8_0 measurement on this machine.
Estimated decode
~76.3–110 t/s
Estimated Q8_0; calculated range
Estimated prefill
~5450–11320 t/s
Estimated Q8_0; calculated range
Estimated TTFT at 8K
~0.8–1.6 s
Estimated Q8_0; calculated range
Hardware load reference
1580 W
Measured machine-level load; not this model run
How the estimate is calculated
- Multi-GPU: 1792 GB/s per card, with each additional card contributing 40% of its bandwidth — 3942 GB/s effective, not the 7168 GB/s aggregate. Cross-GPU collectives use PCIe rather than a dedicated GPU fabric.
- Decode: reading 23.6B active parameters at 8.50 bits/weight takes 9.34 ms at 3942 GB/s x 68% achieved efficiency.
- Prefill: 760 TFLOPS (FP16) x 2 for native FP8 tensor cores x 26% assumed model-FLOPs utilisation, divided by 2 x 23.6B 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.
Submit a benchmarkSubmissions stay pending until reviewed.
This model at other quantizations on this machine
| Quantization | Resident | Disk | Total needed | Utilisation | Fit | Max context |
|---|---|---|---|---|---|---|
Q4_K_M | 13.7 GB | 13.7 GB | 17.6 GB | 5% | Comfortable | 128K |
Q5_K_M | 16.1 GB | 16.1 GB | 20.0 GB | 6% | Comfortable | 128K |
Q8_0recommended | 23.8 GB | 23.8 GB | 28.0 GB | 8% | Comfortable | 128K |
BF16 | 43.9 GB | 43.9 GB | 48.7 GB | 14% | Comfortable | 128K |
Cost of running Devstral Small 24B on this machineUnited States (federal) · C corporation · 8h/day
Monthly economic cost
$746
Calculated after tax
Codex/Claude Code
$100/mo
≈ $100/mo · local is $646 more
Net cash at purchase
$58,553
Calculated VAT not reclaimable
Total over 5 years
$44,773
Calculated after tax, after resale
Cost per USD/1M tokens
$63.26
Calculated 11.8M 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.
Purchase-price input: $58,553 in United States (federal), tax/VAT excluded · estimated · Lenovo enterprise channel (converted catalogue estimate). This is the localized purchase input; the after-tax economic cost is calculated separately below.
Monthly breakdown
| Depreciation 49,770 over 5 years, straight-line to a 8,783 residual | $829.50 |
| Electricity 85.2 kWh/month at 0.140/kWh | $11.93 |
| Cost of capital 4.0%/yr on 33,668 average capital employed | $112.23 |
| Monthly cost before tax | $953.65 |
| Electricity tax shield Running costs are deductible business expenses | −$2.50 |
| First-year expensing §179 (100.0%) | −$204.93 |
| Monthly economic cost after tax | $746.21 |
Three different numbers, deliberately
Cash cost
$58,553
Money that leaves the bank account on day one, net of reclaimable VAT.
Accounting depreciation
$829.50/month
$49,770 written down over 5 years to a $8,783 residual.
After-tax economic cost
$746.21/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
$8,783 (15%) Assumption
Two GPU generations later, the card is worth a fraction of its list price.
Electricity
$0.140/kWh Assumption
Average power draw
484 W Calculated
Load 1580 W for 20% of powered hours, idle 210 W for the rest — a machine that is on is not generating tokens the whole time.
Investment allowances
§179 100.0% Official spec
Worth $12,296 in total — first-year expensing that replaces later tax depreciation.
Cost of capital
$112.23/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
1.6 t/s
Calculated purchase price only
Decode per $100/month
12.5 t/s
Calculated after-tax ownership cost
Tokens per joule
0.19
Calculated same as tokens/s per watt
USD per 1M output tokens
$63.26
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 | 484.6M/mo | $18 | API cheaper | Different model |
| DeepSeek-V4.1 Flash DeepSeek | $0.60 | 414.6M/mo | $21 | API cheaper | Different model |
| DeepSeek-V4 Pro DeepSeek | $1.98 | 102.8M/mo | $86 | API cheaper | Different model |
| DeepSeek V4 Pro (Together) Together AI | $4.40 | 35.2M/mo | $250 | API cheaper | Different model |
| GLM-5.3 Z.ai | $4.40 | 47.8M/mo | $184 | API cheaper | Different model |
| Claude Haiku 4.5 Anthropic | $5.00 | 57.4M/mo | $153 | 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 $746/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.
Devstral Small 24B on other hardware
| Hardware | Decode | Price | Fit |
|---|---|---|---|
| NVIDIA DGX H200 (8x H200, 1,128 GB) | ~354–509 t/s | $359,430 * | Comfortable |
| Quad RTX 5090 workstation (4x 32 GB) | ~87–125 t/s | $12,611 * | Comfortable |
| RTX 5090 workstation (1x 32 GB) | ~65–93.5 t/s | $3,332 * | Comfortable |
| Dual RTX 5090 workstation (2x 32 GB) | ~58.5–84.2 t/s | $5,854 * | Comfortable |
| Dell Precision 7960 Rack (2x RTX PRO 6000, 192 GB) | ~53.7–77.3 t/s | $36,663 * | Comfortable |
| Used RTX 4090 workstation (24 GB) | ~44.3–63.7 t/s | $2,071 * | Comfortable |
Nearest alternatives to the Lenovo ThinkStation PX (4x RTX PRO 6000, 384 GB)
Devstral Small 24B on Dell Precision 7960 Rack (2x RTX PRO 6000, 192 GB)192 GBDevstral Small 24B on NVIDIA DGX H200 (8x H200, 1,128 GB)1,128 GBDevstral Small 24B on NVIDIA DGX Station GB300 (748 GB)748 GBDevstral Small 24B on Mac Studio M5 Ultra 512 GB512 GBDevstral Small 24B on Mac Studio M3 Ultra 512 GB512 GB
Compare Lenovo ThinkStation PX (4x RTX PRO 6000, 384 GB) against Dell Precision 7960 Rack (2x RTX PRO 6000, 192 GB) →
Compare Lenovo ThinkStation PX (4x RTX PRO 6000, 384 GB) against Dell Precision 7960 Rack (2x RTX PRO 6000, 192 GB) →