Created bymario-alka.dePowered bygodcore.denoob2claw.detricoma.de
Contributed byMario AlkaQwen (Alibaba)

Qwen3.8-Flash-Next

Performance benchmark · measured on 29.09.2026 11:10

Benchmark-IDrun-20260929-094617-3e7bb0
Timebench 3 - Kombi (Prefill + Generation)MoE180BRuntime: llama.cppQuantisierung: UD-Q4_K_XL
Generation18,69tok/s
Prefill244,63tok/s
Time to First Token92.958,50ms
Total duration816,48s
Concurrency10parallel
Ranking in the field
59of 59 systems

Performance benchmark · Primary metric: Generation-Speed (tok/s) · 10× concurrent

This run is better than 0 % of all comparable systems.
Generation 18,7 tok/s
-94 % vs Ø 312,6
Prefill 244,6 tok/s
-98 % vs Ø 11.064,5
Time to First Token 92.959 ms
+613 % vs Ø 13.030
Distribution in the field19 – 718 tok/s
Ø 313 Median Dieser Lauf

Wie schlägt sich dieser Benchmark mit anderen Modellen?

How does this benchmark compare on other GPUs?

Same model on different hardware · 10× concurrent · Generation (tok/s)

Hardware

GPU: 2x AMD Radeon RX 7900 XTX · 24 GB VRAM
CPU: AMD Ryzen Threadripper PRO 3955WX 16-Cores
RAM: 189 GB
Mainboard: ASUSTeK COMPUTER INC. Pro WS WRX80E-SAGE SE WIFI

Setup

Runtime: llama.cpp
Quantization: UD-Q4_K_XL
Model: Qwen3.8-Flash-Next

Configuration

benchmark-konfiguration — run-20260929-094617-3e7bb0
# LLM-Benchmark Konfiguration # Modell : Qwen3.8-Flash-Next # Engine : llama.cpp # Run-ID : run-20260929-094617-3e7bb0 # GPU : 2x AMD Radeon RX 7900 XTX # CPU : AMD Ryzen Threadripper PRO 3955WX 16-Cores # RAM : 189 GB bench@llm-benchmark:~$ llama-server Qwen3.8-Flash-Next-UD-Q4_K_XL \ -ngl 999 \ --n-cpu-moe 99 \ -c 24576 \ -np 10 '(2x' RX 7900 XTX + 196GB RAM 'Offload)'
Engine?Die Inferenz-Software, die das Modell ausliefert (z.B. vLLM oder llama.cpp). Sie bestimmt Geschwindigkeit, unterstuetzte Modellformate und welche Parameter ueberhaupt verfuegbar sind.llamacpp
Kontextlaenge?Maximale Anzahl Tokens (Eingabe + erzeugte Ausgabe zusammen), die das Modell pro Anfrage verarbeiten kann.24576
GPU-Layer?Anzahl der auf die GPU ausgelagerten Modell-Layer. Hoeher = mehr VRAM und schneller; der Rest laeuft auf der CPU. 999 = alles auf GPU.999
n-cpu-moe99
Kontext?Groesse des Kontextfensters in Token. 0 = der beim Training verwendete Kontext des Modells.24576
np10
execution_typelocal

All benchmarks of this model To leaderboard

Anzeige
Model comparison

Qwen3.8-Flash-Next on various hardware

All published performance runs of this model – each bubble a variant: position = prefill (X) × generation (Y), bubble size = number of runs. Closer to the top right = faster. ★ Marked gold = this benchmark.

GPUby graphics card

42,334,727,119,411,8151165180194Prefill (tok/s)Generation (tok/s)NVIDIA RTX A6000 - 35,4 tok/s Generation, 165 tok/s Prefill, TTFT 61.921 ms (6 Laufe)NVIDIA RTX A6000AMD Radeon RX 7900 XTX - 18,7 tok/s Generation, 180 tok/s Prefill, TTFT 55.956 ms (3 Laufe) | DIESER LAUF★ AMD Radeon RX 7900 XTX
NVIDIA RTX A6000 35,4 tok/s★ AMD Radeon RX 7900 XTX 18,7 tok/s this run

CPUby processor

42,334,727,119,411,8151165180194Prefill (tok/s)Generation (tok/s)AMD EPYC 7203P 8-Core Processor - 35,4 tok/s Generation, 165 tok/s Prefill, TTFT 61.921 ms (6 Laufe)AMD EPYC 7203P 8-Core...AMD Ryzen Threadripper PRO 3955WX 16-Cores - 18,7 tok/s Generation, 180 tok/s Prefill, TTFT 55.956 ms (3 Laufe) | DIESER LAUF★ AMD Ryzen Threadrippe...
AMD EPYC 7203P 8-Core Processor 35,4 tok/s★ AMD Ryzen Threadripper PRO 3955WX 16-Cores 18,7 tok/s this run

MBby mainboard

39,037,235,433,731,9160167173180Prefill (tok/s)Generation (tok/s)ASUSTeK COMPUTER INC. Pro WS WRX80E-SAGE SE WIFI - 35,4 tok/s Generation, 170 tok/s Prefill, TTFT 59.933 ms (9 Laufe) | DIESER LAUF★ ASUSTeK COMPUTER INC....
★ ASUSTeK COMPUTER INC. Pro WS WRX80E-SAGE SE WIFI 35,4 tok/s this run

ENGby engine

39,037,235,433,731,9160167173180Prefill (tok/s)Generation (tok/s)llama.cpp - 35,4 tok/s Generation, 170 tok/s Prefill, TTFT 59.933 ms (9 Laufe) | DIESER LAUF★ llama.cpp
★ llama.cpp 35,4 tok/s this run

DRVby driver

39,037,235,433,731,9160167173180Prefill (tok/s)Generation (tok/s)unbekannt - 35,4 tok/s Generation, 170 tok/s Prefill, TTFT 59.933 ms (9 Laufe)unbekannt
unbekannt 35,4 tok/s
💰 Economics

Economics of this run

Operating cost, TCO and comparison with the next-best runs of the same model at identical concurrency (10× concurrent). Methodology →

⚙️ ConfigurationAll metrics and charts below follow these settings – based on a 24-month runtime.Save to URLReset
⚡ Electricity price EUR/kWh
⚙️ System utilization 100 %
🖥️ Acquisition EUR
🔌 Idle 10 W
⚡ TDP 720 W
☁️ External LLM (API)
Electricity0.30 EUR/kWh
Avg power (incl. idle)720 W estimated (TDP)GPU 710 + Board 10 W full load
Avg cost / hourEUR 0.22
Electricity / 1M tokensEUR 3.21
Token / kWh93.45K
Acquisition (system)EUR 3,790 partial priceGPU EUR 2,098 · RAM EUR 1,512 · PSU EUR 180
Electricity (2 years)–
TCO (2 years)EUR 7,574
Output tokens (2 years)1.18B
☁️ External LLM (API) – comparison
External LLM cost (2 years)–
Savings vs. external (2 years)–

All values above and the charts below take the configured system utilization into account: at X% the system generates only X% of the time, the rest it idles (10 W). Cost per hour drops (more idle), cost per token rises.

Cost over 2 years – electricity only

Cost over 2 years – incl. acquisition (TCO)

Speed vs. tokens per euro

Euro per 1M tokens

Comparison vs. API – economics per benchmark

Qwen3.8-Flash-Next2x AMD Radeon RX 7900 XTXQwen3.8-Flash-Next2x NVIDIA RTX A6000Qwen3.8-Flash-Next2x NVIDIA RTX A6000
Electricity cost (24 mo.)–––
Acquisition cost–––
Total cost (TCO)–––
Generated tokens (24 mo.)–––
Token price via API–––
Break-even point (days)–––
Result (savings / extra cost)–––

Comparison with up to 3 next-best runs of this model at the same concurrency (at least one on different hardware). Power = GPU TDP + CPU (idle + 15 %) + board (estimated), acquisition = full system (GPU + CPU + board + RAM + PSU), prices = stored market prices.

Contributed by

Mario Alka Administrator

@marioalka

Ich bin Unternehmer, Softwareentwickler und KI-Enthusiast. Seit vielen Jahren entwickle ich Unternehmenssoftware und beschäftige mich inzwischen fast täglich mit lokalen LLMs, KI-Agenten und leistungsfähiger KI-Hardware.

Mit LLM-Benchmark.de möchte ich eine Plattform schaffen, auf der Modelle, GPUs und Agenten objektiv und reproduzierbar miteinander verglichen werden.