Gemma 4 26B-A4B on Mac Studio M5 Ultra 96 GB
25.806B (3.8B active) on 96 GB at 1,200 GB/s. Hardware details · Model details
Yes. Gemma 4 26B-A4B fits on Mac Studio M5 Ultra 96 GB at the recommended
Q8_0 configuration, requiring approximately 29.7 GB at 8K context. Practical context capacity is 192K. Expected decode for the recommended configuration is ~149–214 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
29.7 GB
Calculated of 81.6 GB usable — 36%
Max practical context
192K
Calculated model supports 256K
Memory budget at 8K context
Model weights ⓘ
26.0 GB Calculated
KV cache ⓘ
1.9 GB Calculated
Runtime overhead ⓘ
1.8 GB Estimated
Total required
29.7 GB Calculated
Headroom ⓘ
51.9 GB Calculated
Highest-precision quantization that leaves headroom: uses 36% of usable memory at 8K context.
Unified memory: weights, KV cache and the OS share the same 96 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 25.806B parameters must be resident in memory even though only ~3.8B are active per token. Memory follows total parameters; speed follows active parameters.
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
~149–214 t/s
Estimated Q8_0; calculated range
Estimated prefill
~2340–4870 t/s
Estimated Q8_0; calculated range
Estimated TTFT at 8K
~1.7–3.6 s
Estimated Q8_0; calculated range
Hardware load reference
190 W
Measured machine-level load; not this model run
How the estimate is calculated
- Decode: reading 3.8B active parameters at 8.50 bits/weight takes 3.84 ms at 1200 GB/s x 88% achieved efficiency.
- MoE routing penalty of 15% applied: expert gathers are less bandwidth-efficient than a dense sweep.
- Prefill: 192 TFLOPS (FP16) x 1.69 calibrated against measured prefill on this platform x 22% assumed model-FLOPs utilisation, divided by 2 x 9.9B parameters per token.
- Prefill uses 9.9B effective parameters, not the 3.8B 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.
Submit a benchmarkSubmissions stay pending until reviewed.
This model at other quantizations on this machine
| Quantization | Resident | Disk | Total needed | Utilisation | Fit | Max context |
|---|---|---|---|---|---|---|
MLX 4-bit | 14.1 GB | 14.1 GB | 17.4 GB | 21% | Comfortable | 256K |
Q4_K_M | 15.0 GB | 15.0 GB | 18.3 GB | 22% | Comfortable | 256K |
Q5_K_M | 17.6 GB | 17.6 GB | 21.0 GB | 26% | Comfortable | 256K |
Q8_0recommended | 26.0 GB | 26.0 GB | 29.7 GB | 36% | Comfortable | 192K |
BF16 | 48.1 GB | 48.1 GB | 52.4 GB | 64% | Comfortable | 96K |
Cost of running Gemma 4 26B-A4B on this machineUnited States (federal) · C corporation · 8h/day
Monthly economic cost
$62
Calculated after tax
Codex/Claude Code
$100/mo
≈ $100/mo · local is $38 less
Net cash at purchase
$5,945
Calculated VAT not reclaimable
Total over 5 years
$3,741
Calculated after tax, after resale
Cost per USD/1M tokens
$2.71
Calculated 23.0M 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: $5,945 in United States (federal), tax/VAT excluded · estimated · Apple Store NL (converted catalogue estimate). This is the localized purchase input; the after-tax economic cost is calculated separately below.
Monthly breakdown
| Depreciation 4,161 over 5 years, straight-line to a 1,783 residual | $69.35 |
| Electricity 8.4 kWh/month at 0.140/kWh | $1.17 |
| Cost of capital 4.0%/yr on 3,864 average capital employed | $12.88 |
| Monthly cost before tax | $83.41 |
| Electricity tax shield Running costs are deductible business expenses | −$0.25 |
| First-year expensing §179 (100.0%) | −$20.81 |
| Monthly economic cost after tax | $62.35 |
Three different numbers, deliberately
Cash cost
$5,945
Money that leaves the bank account on day one, net of reclaimable VAT.
Accounting depreciation
$69.35/month
$4,161 written down over 5 years to a $1,783 residual.
After-tax economic cost
$62.35/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
$1,783 (30%) Assumption
Extrapolated from the three-year curve.
Electricity
$0.140/kWh Assumption
Average power draw
48 W Calculated
Load 190 W for 20% of powered hours, idle 12 W for the rest — a machine that is on is not generating tokens the whole time.
Investment allowances
§179 100.0% Official spec
Worth $1,248 in total — first-year expensing that replaces later tax depreciation.
Cost of capital
$12.88/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
30.5 t/s
Calculated purchase price only
Decode per $100/month
290.7 t/s
Calculated after-tax ownership cost
Tokens per joule
3.81
Calculated same as tokens/s per watt
USD per 1M output tokens
$2.71
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 | 40.5M/mo | $35 | API cheaper | Different model |
| DeepSeek-V4.1 Flash DeepSeek | $0.60 | 34.6M/mo | $41 | API cheaper | Different model |
| DeepSeek-V4 Pro DeepSeek | $1.98 | 8.6M/mo | $167 | Local cheaper | Different model |
| DeepSeek V4 Pro (Together) Together AI | $4.40 | 2.9M/mo | $487 | Local cheaper | Different model |
| GLM-5.3 Z.ai | $4.40 | 4.0M/mo | $358 | Local cheaper | Different model |
| Claude Haiku 4.5 Anthropic | $5.00 | 4.8M/mo | $299 | 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 $62/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.
Gemma 4 26B-A4B on other hardware
| Hardware | Decode | Price | Fit |
|---|---|---|---|
| NVIDIA DGX H200 (8x H200, 1,128 GB) | ~521–750 t/s | $359,430 * | Comfortable |
| Quad RTX 5090 workstation (4x 32 GB) | ~281–404 t/s | $12,611 * | Comfortable |
| RTX 5090 workstation (1x 32 GB) | ~270–388 t/s | $3,332 * | Comfortable |
| Lenovo ThinkStation PX (4x RTX PRO 6000, 384 GB) | ~259–372 t/s | $58,553 * | Comfortable |
| Dual RTX 5090 workstation (2x 32 GB) | ~216–311 t/s | $5,854 * | Comfortable |
| Dell Precision 7960 Rack (2x RTX PRO 6000, 192 GB) | ~203–293 t/s | $36,663 * | Comfortable |
Nearest alternatives to the Mac Studio M5 Ultra 96 GB
Gemma 4 26B-A4B on Mac Studio M5 Max 128 GB128 GBGemma 4 26B-A4B on Mac Studio M5 Max 64 GB64 GBGemma 4 26B-A4B on Mac Studio M5 Ultra 256 GB256 GBGemma 4 26B-A4B on Mac Studio M5 Max 36 GB36 GBGemma 4 26B-A4B on Mac Studio M5 Ultra 512 GB512 GB
Compare Mac Studio M5 Ultra 96 GB against Mac Studio M5 Max 128 GB →
Compare Mac Studio M5 Ultra 96 GB against Mac Studio M5 Max 128 GB →