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

Qwen3.5-122B-A10B

Performance benchmark · measured on 18.08.2026 15:32

Benchmark-IDrun-20260818-155046-411ece
Timebench 3 - Kombi (Prefill + Generation)MoE122BRuntime: llama.cppQuantisierung: Q8_0
Generation14,89tok/s
Prefill89,38tok/s
Time to First Token21.613,00ms
Total duration180,79s
Concurrency1parallel
Ranking in the field
76of 81 systems

Performance benchmark · Primary metric: Generation-Speed (tok/s) · 1× concurrent

This run is better than 6 % of all comparable systems.
Generation 14,9 tok/s
-72 % vs Ø 53,3
Prefill 89,4 tok/s
-94 % vs Ø 1.382,6
Time to First Token 21.613 ms
+411 % vs Ø 4.232
Distribution in the field10 – 155 tok/s
Ø 53 Median Dieser Lauf

Wie schlägt sich dieser Benchmark mit anderen Modellen?

How does this benchmark compare on other GPUs?

Same model on different hardware · 1× concurrent · Generation (tok/s)

Hardware

GPU: 2x AMD Radeon PRO W7900 Dual Slot · 48 GB VRAM
CPU: AMD Ryzen Threadripper PRO 5975WX 32-Cores
RAM: 247 GB
Mainboard: ASUSTeK COMPUTER INC. Pro WS WRX80E-SAGE SE WIFI

Setup

Runtime: llama.cpp
Quantization: Q8_0
Model: Qwen3.5-122B-A10B

Configuration

benchmark-konfiguration — run-20260818-155046-411ece
# LLM-Benchmark Konfiguration # Modell : Qwen3.5-122B-A10B # Engine : llama.cpp # Run-ID : run-20260818-155046-411ece # GPU : 2x AMD Radeon PRO W7900 Dual Slot # CPU : AMD Ryzen Threadripper PRO 5975WX 32-Cores # RAM : 247 GB bench@llm-benchmark:~$ /home/godcore/llama.cpp/build_hip/bin/llama-server \ -m /home/godcore/bench_scratch/Qwen_Qwen3.5-122B-A10B-Q8_0-00001-of-00004.gguf \ --alias Qwen3.5-122B-A10B \ -ngl 999 \ -fa on \ -sm layer \ --tensor-split 1,1 \ -cmoe \ -c 49152 \ -np 12 \ --jinja \ --host 0.0.0.0 \ --port 8000
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
Modellalias?Der Name, unter dem das Modell ueber die API angesprochen wird. Genau dieser Wert muss im Request-Feld 'model' stehen.Qwen3.5-122B-A10B
Kontextlaenge?Maximale Anzahl Tokens (Eingabe + erzeugte Ausgabe zusammen), die das Modell pro Anfrage verarbeiten kann.49152
Modellpfad?Pfad zur GGUF-Modelldatei, die geladen und ausgeliefert wird./home/godcore/bench_scratch/Qwen_Qwen3.5-122B-A10B-Q8_0-00001-of-00004.gguf
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
faon
Split-Mode?Verteilung ueber mehrere GPUs: none (nur eine GPU), layer (Layer aufteilen) oder row (Tensoren zeilenweise).layer
Tensor-Split?Verhaeltnis, in dem die Modell-Layer auf mehrere GPUs verteilt werden, z.B. 3,1.1,1
Kontext?Groesse des Kontextfensters in Token. 0 = der beim Training verwendete Kontext des Modells.49152
np12

All benchmarks of this model To leaderboard

Anzeige
Model comparison

Qwen3.5-122B-A10B 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

7325493661830,001.2912.5823.873Prefill (tok/s)Generation (tok/s)NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition - 564,6 tok/s Generation, 2.824 tok/s Prefill, TTFT 9.370 ms (12 Laufe)NVIDIA RTX PRO 6000 B...AMD Radeon AI PRO R9700 - 168,6 tok/s Generation, 3.133 tok/s Prefill, TTFT 24.492 ms (9 Laufe)AMD Radeon AI PRO R97...NVIDIA GeForce RTX 5070 Ti - 31,1 tok/s Generation, 94 tok/s Prefill, TTFT 168.001 ms (8 Laufe)NVIDIA GeForce RTX 50...NVIDIA GeForce RTX 3090 Ti - 24,9 tok/s Generation, 129 tok/s Prefill, TTFT 137.609 ms (3 Laufe)NVIDIA GeForce RTX 30...NVIDIA RTX PRO 6000 Blackwell Workstation Edition - 12,1 tok/s Generation, 71 tok/s Prefill, TTFT 105.658 ms (5 Laufe)NVIDIA RTX PRO 6000 B...NVIDIA GeForce RTX 5090 - 8,3 tok/s Generation, 72 tok/s Prefill, TTFT 99.190 ms (8 Laufe)NVIDIA GeForce RTX 50...AMD Radeon PRO W7900 Dual Slot - 96,5 tok/s Generation, 420 tok/s Prefill, TTFT 52.568 ms (9 Laufe) | DIESER LAUF★ AMD Radeon PRO W7900 ...
NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition 564,6 tok/sAMD Radeon AI PRO R9700 168,6 tok/s★ AMD Radeon PRO W7900 Dual Slot 96,5 tok/s this runNVIDIA GeForce RTX 5070 Ti 31,1 tok/sNVIDIA GeForce RTX 3090 Ti 24,9 tok/sNVIDIA RTX PRO 6000 Blackwell Workstation Edition 12,1 tok/sNVIDIA GeForce RTX 5090 8,3 tok/s

CPUby processor

7325493661830,001.2912.5823.873Prefill (tok/s)Generation (tok/s)AMD Ryzen Threadripper PRO 9965WX 24-Cores - 564,6 tok/s Generation, 2.824 tok/s Prefill, TTFT 9.370 ms (12 Laufe)AMD Ryzen Threadrippe...AMD Ryzen Threadripper PRO 7955WX 16-Cores - 168,6 tok/s Generation, 3.133 tok/s Prefill, TTFT 24.492 ms (9 Laufe)AMD Ryzen Threadrippe...AMD Ryzen 9 8945HX with Radeon Graphics - 24,9 tok/s Generation, 129 tok/s Prefill, TTFT 137.609 ms (3 Laufe)AMD Ryzen 9 8945HX wi...AMD Ryzen 9 9950X 16-Core Processor - 12,1 tok/s Generation, 71 tok/s Prefill, TTFT 105.658 ms (5 Laufe)AMD Ryzen 9 9950X 16-...AMD Ryzen 7 5800X3D 8-Core Processor - 8,3 tok/s Generation, 72 tok/s Prefill, TTFT 99.190 ms (8 Laufe)AMD Ryzen 7 5800X3D 8...AMD Ryzen Threadripper PRO 5975WX 32-Cores - 96,5 tok/s Generation, 266 tok/s Prefill, TTFT 106.889 ms (17 Laufe) | DIESER LAUF★ AMD Ryzen Threadrippe...
AMD Ryzen Threadripper PRO 9965WX 24-Cores 564,6 tok/sAMD Ryzen Threadripper PRO 7955WX 16-Cores 168,6 tok/s★ AMD Ryzen Threadripper PRO 5975WX 32-Cores 96,5 tok/s this runAMD Ryzen 9 8945HX with Radeon Graphics 24,9 tok/sAMD Ryzen 9 9950X 16-Core Processor 12,1 tok/sAMD Ryzen 7 5800X3D 8-Core Processor 8,3 tok/s

MBby mainboard

7325493661830,001.2182.4353.653Prefill (tok/s)Generation (tok/s)ASUSTeK COMPUTER INC. Pro WS WRX90E-SAGE SE - 564,6 tok/s Generation, 2.956 tok/s Prefill, TTFT 15.851 ms (21 Laufe)ASUSTeK COMPUTER INC....Meigao Innovation Technology (Shen Zhen) Co., Ltd DRFXL (MotherBoard Series) - 24,9 tok/s Generation, 129 tok/s Prefill, TTFT 137.609 ms (3 Laufe)Meigao Innovation Tec...ASUSTeK COMPUTER INC. ProArt X870E-CREATOR WIFI - 12,1 tok/s Generation, 71 tok/s Prefill, TTFT 105.658 ms (5 Laufe)ASUSTeK COMPUTER INC....ASUSTeK COMPUTER INC. ROG STRIX B550-A GAMING - 8,3 tok/s Generation, 72 tok/s Prefill, TTFT 99.190 ms (8 Laufe)ASUSTeK COMPUTER INC....ASUSTeK COMPUTER INC. Pro WS WRX80E-SAGE SE WIFI - 96,5 tok/s Generation, 266 tok/s Prefill, TTFT 106.889 ms (17 Laufe) | DIESER LAUF★ ASUSTeK COMPUTER INC....
ASUSTeK COMPUTER INC. Pro WS WRX90E-SAGE SE 564,6 tok/s★ ASUSTeK COMPUTER INC. Pro WS WRX80E-SAGE SE WIFI 96,5 tok/s this runMeigao Innovation Technology (Shen Zhen) Co., Ltd DRFXL (MotherBoard Series) 24,9 tok/sASUSTeK COMPUTER INC. ProArt X870E-CREATOR WIFI 12,1 tok/sASUSTeK COMPUTER INC. ROG STRIX B550-A GAMING 8,3 tok/s

ENGby engine

70053336720032,903.7787.55611.334Prefill (tok/s)Generation (tok/s)vLLM - 168,6 tok/s Generation, 9.255 tok/s Prefill, TTFT 1.415 ms (3 Laufe)vLLMllama.cpp - 564,6 tok/s Generation, 788 tok/s Prefill, TTFT 76.086 ms (51 Laufe) | DIESER LAUF★ llama.cpp
★ llama.cpp 564,6 tok/s this runvLLM 168,6 tok/s

DRVby driver

6215935655365081.1821.2331.2831.333Prefill (tok/s)Generation (tok/s)unbekannt - 564,6 tok/s Generation, 1.258 tok/s Prefill, TTFT 71.937 ms (54 Laufe)unbekannt
unbekannt 564,6 tok/s
💰 Economics

Economics of this run

Operating cost, TCO and comparison with the next-best runs of the same model at identical concurrency (1× 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 600 W
☁️ External LLM (API)
Electricity0.30 EUR/kWh
Avg power (incl. idle)600 W estimated (TDP)GPU 590 + Board 10 W full load
Avg cost / hourEUR 0.18
Electricity / 1M tokensEUR 3.36
Token / kWh89.34K
Acquisition (system)EUR 2,156 missingRAM EUR 1,976 · PSU EUR 180
Electricity (2 years)
TCO (2 years)EUR 5,310
Output tokens (2 years)939.14M
☁️ 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

Qwen3.5-122B-A10B2x AMD Radeon PRO W7900 Dual SlotQwen3.5-122B-A10B3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation EditionQwen3.5-122B-A10B3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation EditionQwen3.5-122B-A10B3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition
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.