Mistral Small 3.2 24B on Mac Studio M5 Max 36 GB

24.011B on 36 GB at 614 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
Q4_K_M
Calculated gguf
Memory required
16.6 GB
Calculated of 30.6 GB usable — 54%
Max practical context
64K
Calculated model supports 128K

Memory budget at 8K context

Model weights
14.0 GB Calculated
KV cache
1.3 GB Calculated
Runtime overhead
1.4 GB Estimated
Total required
16.6 GB Calculated
Usable memory
30.6 GB Assumption
Headroom
14.0 GB Calculated
Highest-precision quantization that leaves headroom: uses 54% of usable memory at 8K context.
Unified memory: weights, KV cache and the OS share the same 36 GB pool. We assume 85% is addressable by the GPU, which matches a headless machine with the wired-memory limit raised.
PerformanceEstimated0/10
Decode (generation)
~29.2–42.1 t/s
Estimated calculated, not measured
Prefill (prompt)
~483–1000 t/s
Estimated
TTFT at 8K
~8.3–17.1 s
Estimated time to first token
Power while generating
105 W
Measured 1.34 tokens/s per watt
These figures are estimated, not measured. Nobody has published a benchmark of Mistral Small 3.2 24B 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 24.0B active parameters at 4.85 bits/weight takes 27.05 ms at 614 GB/s x 88% achieved efficiency.
  2. Prefill: 96 TFLOPS (FP16) x 1.69 calibrated against measured prefill on this platform x 22% assumed model-FLOPs utilisation, divided by 2 x 24.0B 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-bit13.1 GB15.7 GB51%Comfortable64K
Q4_K_Mrecommended14.0 GB16.6 GB54%Comfortable64K
Q8_024.2 GB27.2 GB89%Fits8K
BF1644.7 GB48.3 GB158%Does not fit0
Cost of running Mistral Small 3.2 24B on this machineUnited States (federal) · C corporation · 8h/day
Monthly economic cost
$28
Calculated after tax
Codex/Claude Code
$100/mo
≈ $100/mo · local is $72 less
Net cash at purchase
$2,702
Calculated VAT not reclaimable
Total over 5 years
$1,706
Calculated after tax, after resale
Cost per USD/1M tokens
$6.30
Calculated 4.5M 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
1,891 over 5 years, straight-line to a 810 residual
$31.52
Electricity
4.7 kWh/month at 0.140/kWh
$0.66
Cost of capital
4.0%/yr on 1,756 average capital employed
$5.85
Monthly cost before tax$38.03
Electricity tax shield
Running costs are deductible business expenses
−$0.14
First-year expensing
§179 (100.0%)
−$9.46
Monthly economic cost after tax$28.43

Three different numbers, deliberately

Cash cost
$2,702
Money that leaves the bank account on day one, net of reclaimable VAT.
Accounting depreciation
$31.52/month
$1,891 written down over 5 years to a $810 residual.
After-tax economic cost
$28.43/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
$810 (30%) Assumption
Extrapolated from the three-year curve.
Electricity
$0.140/kWh Assumption
Average power draw
27 W Calculated
Load 105 W for 20% of powered hours, idle 7 W for the rest — a machine that is on is not generating tokens the whole time.
Investment allowances
§179 100.0% Official spec
Worth $567 in total — first-year expensing that replaces later tax depreciation.
Cost of capital
$5.85/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
13.2 t/s
Calculated purchase price only
Decode per $100/month
125.4 t/s
Calculated after-tax ownership cost
Tokens per joule
1.34
Calculated same as tokens/s per watt
USD per 1M output tokens
$6.30
Calculated at 20% utilisation
Local versus hosted APIs
Hosted modelUSD/1M outputBreak-evenAPI at your volumeVerdictComparison type
Qwen3.8 Flash
Alibaba Cloud
$0.4218.5M/mo$7API cheaperDifferent model
DeepSeek-V4 Flash
DeepSeek
$0.6611.7M/mo$11API cheaperDifferent model
DeepSeek-V4 Pro
DeepSeek
$1.983.9M/mo$33Local cheaperDifferent model
DeepSeek V4 Pro (Together)
Together AI
$4.401.3M/mo$96Local cheaperDifferent model
GLM-5.3
Z.ai
$4.401.8M/mo$70Local cheaperDifferent model
Claude Haiku 4.5
Anthropic
$5.002.2M/mo$59Local 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 $28/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.
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)