Qwen3.8 Flash Next on GMKtec EVO-X2 (Ryzen AI Max+ 395, 128 GB)
180B (6B active) on 128 GB at 256 GB/s. Hardware details · Model details
CompatibilityDoes not fit
Fits
No
Calculated Requires more than 97% of usable memory. Needs a smaller quantization or more memory.
Recommended quantization
MLX 4-bit
Calculated mlx
Memory required
102.8 GB
Calculated of 105.0 GB usable — 98%
Max practical context
0
Calculated model supports 256K
Memory budget at 8K context
Model weights ⓘ
98.1 GB Calculated
KV cache ⓘ
0.8 GB Calculated
Runtime overhead ⓘ
3.9 GB Estimated
Total required
102.8 GB Calculated
Usable memory ⓘ
105.0 GB Assumption
Headroom
2.2 GB Calculated
No quantization in our catalogue fits this machine.
Unified LPDDR: the accelerator allocates from the same 128 GB pool as the OS. We assume 82% is available to the inference engine.
Mixture of experts: all 180B parameters must be resident in memory even though only ~6B are active per token. Memory follows total parameters; speed follows active parameters.
PerformanceEstimated0/10
Decode (generation)
—
Estimated calculated, not measured
Prefill (prompt)
—
Estimated
TTFT at 8K
~37.6–78.1 s
Estimated time to first token
Power while generating
135 W
Measured
No throughput figures for this pairing. We do not estimate performance for a model that cannot load: a tokens-per-second number for a configuration that will never run is noise, not data. The alternatives below are machines that can actually run Qwen3.8 Flash Next.
How the estimate is calculated
- Decode: reading 6.0B active parameters at 4.50 bits/weight takes 31.50 ms at 256 GB/s x 42% achieved efficiency.
- MoE routing penalty of 15% applied: expert gathers are less bandwidth-efficient than a dense sweep.
- Prefill: 59 TFLOPS (FP16) x 18% assumed model-FLOPs utilisation, divided by 2 x 32.9B parameters per token.
- Prefill uses 32.9B effective parameters, not the 6B 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-bitrecommended | 98.1 GB | 102.8 GB | 98% | Does not fit | 0 |
Q4_K_M | 104.7 GB | 109.6 GB | 104% | Does not fit | 0 |
Q8_0 | 181.7 GB | 188.9 GB | 180% | Does not fit | 0 |
BF16 | 335.3 GB | 347.1 GB | 331% | Does not fit | 0 |
Cost of running Qwen3.8 Flash Next on this machineUnited States (federal) · C corporation · 8h/day
Monthly economic cost
$22
Calculated after tax
Codex/Claude Code
$100/mo
≈ $100/mo · local is $78 less
Net cash at purchase
$1,711
Calculated VAT not reclaimable
Total over 5 years
$1,336
Calculated after tax, after resale
Cost per USD/1M tokens
—
Calculated
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 1,454 over 5 years, straight-line to a 257 residual | $24.23 |
| Electricity 6.7 kWh/month at 0.140/kWh | $0.94 |
| Cost of capital 4.0%/yr on 984 average capital employed | $3.28 |
| Monthly cost before tax | $28.45 |
| Electricity tax shield Running costs are deductible business expenses | −$0.20 |
| First-year expensing §179 (100.0%) | −$5.99 |
| Monthly economic cost after tax | $22.27 |
Three different numbers, deliberately
Cash cost
$1,711
Money that leaves the bank account on day one, net of reclaimable VAT.
Accounting depreciation
$24.23/month
$1,454 written down over 5 years to a $257 residual.
After-tax economic cost
$22.27/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
$257 (15%) Assumption
Two GPU generations later, the card is worth a fraction of its list price.
Electricity
$0.140/kWh Assumption
Average power draw
38 W Calculated
Load 135 W for 20% of powered hours, idle 14 W for the rest — a machine that is on is not generating tokens the whole time.
Investment allowances
§179 100.0% Official spec
Worth $359 in total — first-year expensing that replaces later tax depreciation.
Cost of capital
$3.28/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
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.
Qwen3.8 Flash Next on other hardware
| Hardware | Decode | Price | Fit |
|---|---|---|---|
| NVIDIA DGX H200 (8x H200, 1,128 GB) | ~490–705 t/s | $359,430 * | Comfortable |
| Quad RTX 5090 workstation (4x 32 GB) | ~307–442 t/s | $12,611 * | Borderline |
| Dell Precision 7960 Rack (2x RTX PRO 6000, 192 GB) | ~218–313 t/s | $36,663 * | Comfortable |
| Lenovo ThinkStation PX (4x RTX PRO 6000, 384 GB) | ~195–281 t/s | $58,553 * | Comfortable |
| Mac Studio M5 Ultra 256 GB | ~162–233 t/s | $8,647 * | Comfortable |
| NVIDIA DGX Station GB300 (748 GB) | ~102–147 t/s | $90,082 * | Comfortable |
Nearest alternatives to the GMKtec EVO-X2 (Ryzen AI Max+ 395, 128 GB)
Qwen3.8 Flash Next on Framework Desktop (Ryzen AI Max+ 395, 128 GB)128 GBQwen3.8 Flash Next on NVIDIA DGX Spark 128 GB128 GBQwen3.8 Flash Next on CPU-only workstation (Ryzen 9950X, 192 GB DDR5)192 GBQwen3.8 Flash Next on Mac Studio M5 Max 128 GB128 GBQwen3.8 Flash Next on Radeon AI PRO R9700 workstation (32 GB)32 GB
Compare GMKtec EVO-X2 (Ryzen AI Max+ 395, 128 GB) against Framework Desktop (Ryzen AI Max+ 395, 128 GB) →
Compare GMKtec EVO-X2 (Ryzen AI Max+ 395, 128 GB) against Framework Desktop (Ryzen AI Max+ 395, 128 GB) →
Page quality score (why this page is or is not indexed)
6/13 — marked noindex. 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 price (+2)
- Has an ownership economics calculation (+2)
- Well connected (18 internal links) (+1)
- No benchmark or estimate
- Model does not fit and there is no measurement to explain why it is worth knowing (-2)
- Fails a hard requirement: an indexable page needs both a compatibility calculation and at least one benchmark or estimate.