Contributed byMario AlkaQwen (Alibaba) Qwen2.5-Coder-32B-Instruct
Performance benchmark · measured on 21.07.2026 10:30
Benchmark-ID
run-20260722-165906-348187Timebench 3 - Kombi (Prefill + Generation)Dense32BRuntime: llama.cppQuantisierung: Q4_K_M
Generation19,08tok/s
Prefill1.508,31tok/s
Time to First Token1.441,00ms
Total duration55,12s
Concurrency1parallel
Ranking in the field
1177of 1549 systems
Performance benchmark · Primary metric: Generation-Speed (tok/s) · 1× concurrent
This run is better than 24 % of all comparable systems.
Generation
19,1 tok/s
-76 % vs Ø 79,0
Prefill
1.508,3 tok/s
-47 % vs Ø 2.837,1
Time to First Token
1.441 ms
-95 % vs Ø 26.963
Distribution in the field0 – 405 tok/s
Wie schlägt sich dieser Benchmark mit anderen Modellen?
gpt-oss-20bNVIDIA GeForce RTX 5090 · run-20260724-004608-16af9b
404,6 tok/s
gpt-oss-20bNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260730-035052-9f766a
395,2 tok/s
Nemotron-3-Nano-4BNVIDIA GeForce RTX 5090 · run-20260728-194135-d456e7
393,5 tok/s
Nemotron-3-Nano-4BNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260727-181607-bffb11
393,5 tok/s
gpt-oss-20bNVIDIA GeForce RTX 5090 · run-20260724-004607-29bd16
388,9 tok/s
gpt-oss-20bNVIDIA GeForce RTX 5090 · run-20260729-032121-1c779f
388,2 tok/s
gemma-4-E2B-itNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260727-181606-6ea089
378,6 tok/s
gemma-4-E2B-itNVIDIA GeForce RTX 5090 · run-20260728-184455-e8b129
377,6 tok/s
Nemotron-3-Nano-4B3× NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260728-194136-ac388e
362,3 tok/s
Nemotron-3-Nano-4B3× NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260728-194136-5e5085
361,4 tok/s
Nemotron-3-Nano-30B-A3BNVIDIA GeForce RTX 5090 · run-20260729-105335-653788
359,4 tok/s
Nemotron-3-Nano-Omni-30B-A3B-ReasoningNVIDIA GeForce RTX 5090 · run-20260729-105334-2c4905
359,3 tok/s
Nemotron-Cascade-2-30B-A3BNVIDIA GeForce RTX 5090 · run-20260729-105334-44d1e7
359,0 tok/s
Nemotron-3-Nano-Omni-30B-A3B-ReasoningNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260729-032119-4852bf
357,1 tok/s
Qwen2.5-Coder-32B-Instruct this run3× AMD Radeon AI PRO R9700 · run-20260722-165906-348187
19,1 tok/s
How does this benchmark compare on other GPUs?
Same model on different hardware · 1× concurrent · Generation (tok/s)
NVIDIA RTX PRO 6000 Blackwell Workstation Editionrun-20260724-082952-475bf4
70,2 tok/s
NVIDIA GeForce RTX 5090run-20260730-181639-7b083a
68,8 tok/s
NVIDIA RTX PRO 6000 Blackwell Workstation Editionrun-20260729-032127-a4486e
68,3 tok/s
3× NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Editionrun-20260730-181641-b834fb
64,6 tok/s
3× NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Editionrun-20260730-181642-6e8860
64,5 tok/s
2× NVIDIA RTX A6000run-20260829-131336-ee6d22
58,5 tok/s
3× NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Editionrun-20260730-181641-161671
57,1 tok/s
2× NVIDIA RTX A6000run-20260827-094103-4bee57
54,7 tok/s
NVIDIA GeForce RTX 3090 Tirun-20260729-105334-d27d6d
40,2 tok/s
2× NVIDIA RTX A6000run-20260829-220129-a69899
36,0 tok/s
2× NVIDIA RTX A6000run-20260827-161406-c61e3d
34,6 tok/s
NVIDIA RTX A6000run-20260826-215014-23f59d
30,3 tok/s
2× AMD Radeon PRO W7900 Dual Slotrun-20260818-122245-5babb9
27,7 tok/s
AMD Radeon PRO W7900 Dual Slotrun-20260817-142935-1fb150
27,6 tok/s
Hardware
GPU: 3x AMD Radeon AI PRO R9700 · 32 GB VRAM
CPU: AMD Ryzen Threadripper PRO 7955WX 16-Cores
RAM: 184 GB
Mainboard: ASUSTeK COMPUTER INC. Pro WS WRX90E-SAGE SE
Setup
Runtime: llama.cpp
Quantization: Q4_K_M
Model: Qwen2.5-Coder-32B-Instruct
Configuration
# LLM-Benchmark Konfiguration
# Modell : Qwen2.5-Coder-32B-Instruct
# Engine : llama.cpp
# Run-ID : run-20260722-165906-348187
# GPU : 3x AMD Radeon AI PRO R9700
# CPU : AMD Ryzen Threadripper PRO 7955WX 16-Cores
# RAM : 184 GB
bench@llm-benchmark:~$ /src/build/bin/llama-server \
-m /root/.cache/huggingface/hub/models--bartowski--Qwen2.5-Coder-32B-Instruct-GGUF/snapshots/40b525506a4f98ed425882fa6dfc90cc8139065e/Qwen2.5-Coder-32B-Instruct-Q4_K_M.gguf \
-a Qwen2.5-Coder-32B-Instruct \
-ngl 999 \
--split-mode layer \
--host 0.0.0.0 \
--port 8000 \
-c 8192 \
--no-webui
| 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. | Qwen2.5-Coder-32B-Instruct |
| Kontextlaenge?Maximale Anzahl Tokens (Eingabe + erzeugte Ausgabe zusammen), die das Modell pro Anfrage verarbeiten kann. | 8192 |
| Modellpfad?Pfad zur GGUF-Modelldatei, die geladen und ausgeliefert wird. | /root/.cache/huggingface/hub/models--bartowski--Qwen2.5-Coder-32B-Instruct-GGUF/snapshots/40b525506a4f98ed425882fa6dfc90cc8139065e/Qwen2.5-Coder-32B-Instruct-Q4_K_M.gguf |
| Alias?Anzeigename des Modells nach aussen (served model name), unabhaengig vom Dateinamen. | Qwen2.5-Coder-32B-Instruct |
| 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 |
| Split-Mode?Verteilung ueber mehrere GPUs: none (nur eine GPU), layer (Layer aufteilen) oder row (Tensoren zeilenweise). | layer |
| Kontext?Groesse des Kontextfensters in Token. 0 = der beim Training verwendete Kontext des Modells. | 8192 |
| Web-UI aus?Deaktiviert die eingebaute Web-Oberflaeche; nur die OpenAI-kompatible API bleibt erreichbar. | aktiv |
Anzeige
Model comparison
Qwen2.5-Coder-32B-Instruct 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
NVIDIA RTX PRO 6000 Blackwell Workstation Edition 500,6 tok/sNVIDIA GeForce RTX 5090 479,6 tok/sNVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition 412,8 tok/sNVIDIA RTX A6000 362,6 tok/sNVIDIA GeForce RTX 3090 Ti 236,4 tok/sAMD Radeon PRO W7900 Dual Slot 102,1 tok/sAMD Radeon PRO W7800 48GB 69,8 tok/sNVIDIA GeForce RTX 5070 Ti 45,3 tok/s★ AMD Radeon AI PRO R9700 21,2 tok/s this run
CPUby processor
AMD Ryzen 9 9950X 16-Core Processor 500,6 tok/sAMD Ryzen 7 5800X3D 8-Core Processor 479,6 tok/sAMD Ryzen Threadripper PRO 9965WX 24-Cores 412,8 tok/s★ AMD Ryzen Threadripper PRO 7955WX 16-Cores 362,6 tok/s this runAMD Ryzen 9 8945HX with Radeon Graphics 236,4 tok/sAMD Ryzen Threadripper PRO 5975WX 32-Cores 102,1 tok/s
MBby mainboard
ASUSTeK COMPUTER INC. ProArt X870E-CREATOR WIFI 500,6 tok/sASUSTeK COMPUTER INC. ROG STRIX B550-A GAMING 479,6 tok/s★ ASUSTeK COMPUTER INC. Pro WS WRX90E-SAGE SE 412,8 tok/s this runMeigao Innovation Technology (Shen Zhen) Co., Ltd DRFXL (MotherBoard Series) 236,4 tok/sASUSTeK COMPUTER INC. Pro WS WRX80E-SAGE SE WIFI 102,1 tok/s
DRVby driver
unbekannt 500,6 tok/sAMD 7.0.0-27-generic 21,2 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
125 W
⚡ TDP
971 W
☁️ External LLM (API)
Electricity0.30 EUR/kWh
Avg power (incl. idle)971 W estimated (TDP)GPU 900 + CPU 46 + Board 25 W full load
Avg cost / hourEUR 0.29
Electricity / 1M tokensEUR 4.24
Token / kWh70.74K
Acquisition (system)EUR 9,674 full priceGPU EUR 4,200 · CPU EUR 1,399 · Board EUR 1,299 · RAM EUR 2,576 · PSU EUR 200
Electricity (2 years)–
TCO (2 years)EUR 14,778
Output tokens (2 years)1.20B
☁️ 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 (125 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
| Qwen2.5-Coder-32B-Instruct3x AMD Radeon AI PRO R9700 | Qwen2.5-Coder-32B-InstructNVIDIA RTX PRO 6000 Blackwell Workstation Edition | Qwen2.5-Coder-32B-InstructNVIDIA GeForce RTX 5090 | Qwen2.5-Coder-32B-InstructNVIDIA RTX PRO 6000 Blackwell 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) | – | – | – | – |
3× AMD Radeon AI PRO R9700 · this run 19 tok/sNVIDIA RTX PRO 6000 Blackwell Workstation Edition 70 tok/sNVIDIA GeForce RTX 5090 69 tok/sNVIDIA RTX PRO 6000 Blackwell Workstation Edition 68 tok/s
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.