Contributed byMario AlkaNVIDIA OpenReasoning-Nemotron-32B
Performance benchmark · measured on 23.07.2026 14:11
Benchmark-ID
run-20260723-142108-cc471cTimebench 3 - Kombi (Prefill + Generation)Dense32BRuntime: llama.cppQuantisierung: Q4_K_M
ℹ
For context: Diese Plattform nutzt Unified Memory – die "VRAM" ist gemeinsamer System-RAM (APU/Superchip); das Modell teilt sich den Speicher mit dem System.
Generation16,53tok/s
Prefill691,28tok/s
Time to First Token41.832,00ms
Total duration124,85s
Concurrency10parallel
Ranking in the field
13of 13 systems
Performance benchmark · Primary metric: Generation-Speed (tok/s) · 10× concurrent
This run is better than 0 % of all comparable systems.
Generation
16,5 tok/s
-79 % vs Ø 80,0
Prefill
691,3 tok/s
-48 % vs Ø 1.327,7
Time to First Token
41.832 ms
+96 % vs Ø 21.349
Distribution in the field17 – 158 tok/s
Wie schlägt sich dieser Benchmark mit anderen Modellen?
Kolibri-1AMD Radeon 8060S Graphics · run-20261008-165122-05f32b
158,2 tok/s
North-Mini-Code-1.0AMD Radeon 8060S Graphics · run-20260723-015518-753958
131,9 tok/s
Qwen3-30B-A3B-Instruct-2507AMD Radeon 8060S Graphics · run-20260723-015518-4f6b2e
107,6 tok/s
Qwen-AgentWorld-35B-A3BAMD Radeon 8060S Graphics · run-20260723-142109-66f9ec
103,8 tok/s
Nemotron-Cascade-2-30B-A3BAMD Radeon 8060S Graphics · run-20260723-142105-4bf79a
94,6 tok/s
Gemma-4-26B-A4B-itAMD Radeon 8060S Graphics · run-20260725-031733-4e4df6
76,3 tok/s
gemma-4-26B-A4B-it-assistantAMD Radeon 8060S Graphics · run-20260725-031727-c6550e
74,5 tok/s
gpt-oss-20bAMD Radeon 8060S Graphics · run-20260724-082941-665f99
74,4 tok/s
Gemma-4-26B-A4B-itAMD Radeon 8060S Graphics · run-20260725-031733-e7513a
73,7 tok/s
Ministral-3-14B-Reasoning-2512AMD Radeon 8060S Graphics · run-20260723-142059-0f740f
66,3 tok/s
Qwen3.6-27BAMD Radeon 8060S Graphics · run-20260723-194446-8d651c
36,1 tok/s
Mistral-Small-3.1-24B-Instruct-2503AMD Radeon 8060S Graphics · run-20260723-142100-b7b090
26,0 tok/s
OpenReasoning-Nemotron-32B this runAMD Radeon 8060S Graphics · run-20260723-142108-cc471c
16,5 tok/s
How does this benchmark compare on other GPUs?
Same model on different hardware · 10× concurrent · Generation (tok/s)
NVIDIA GeForce RTX 5090run-20260730-084906-a113b8
481,0 tok/s
NVIDIA RTX PRO 6000 Blackwell Workstation Editionrun-20260729-032126-5eacfa
480,0 tok/s
3× NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Editionrun-20260730-084907-ba8797
411,7 tok/s
3× NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Editionrun-20260730-084907-c82f89
391,8 tok/s
3× NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Editionrun-20260730-084907-194520
342,6 tok/s
NVIDIA GeForce RTX 3090 Tirun-20260729-105332-81a03d
236,8 tok/s
2× NVIDIA RTX A6000run-20260829-223421-8122c2
144,0 tok/s
2× NVIDIA RTX A6000run-20260828-101910-30d7c7
140,0 tok/s
2× NVIDIA RTX A6000run-20260830-043727-dee994
138,4 tok/s
2× NVIDIA RTX A6000run-20260828-080811-cc9cc5
137,8 tok/s
2× AMD Radeon RX 7900 XTXrun-20261009-081231-5b2e57
124,6 tok/s
AMD Radeon AI PRO R9700run-20260922-112426-565e9e
116,5 tok/s
NVIDIA RTX A6000run-20260826-145351-8b2a5a
115,9 tok/s
NVIDIA GeForce RTX 5070 Tirun-20260729-032132-e6debb
49,9 tok/s
Hardware
GPU: AMD Radeon 8060S Graphics
CPU: AMD RYZEN AI MAX+ 395 w/ Radeon 8060S
RAM: 31 GB
Mainboard: Meigao Innovation Technology (Shen Zhen) Co., Ltd SHWSA (MS-S1 MAX)
Setup
Runtime: llama.cpp
Quantization: Q4_K_M
Model: OpenReasoning-Nemotron-32B
Configuration
# LLM-Benchmark Konfiguration
# Modell : OpenReasoning-Nemotron-32B
# Engine : llama.cpp
# Run-ID : run-20260723-142108-cc471c
# GPU : AMD Radeon 8060S Graphics
# CPU : AMD RYZEN AI MAX+ 395 w/ Radeon 8060S
# RAM : 31 GB
bench@llm-benchmark:~$ /home/godcore/llama.cpp/build-rocm/bin/llama-server \
-m /home/godcore/models/dl/nvidia_OpenReasoning-Nemotron-32B-Q4_K_M.gguf \
--host 0.0.0.0 \
--port 8000 \
--gpu-layers 99 \
--flash-attn on \
-c 26624 \
--parallel 12
| 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. | 26624 |
| Modellpfad?Pfad zur GGUF-Modelldatei, die geladen und ausgeliefert wird. | /home/godcore/models/dl/nvidia_OpenReasoning-Nemotron-32B-Q4_K_M.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. | 99 |
| Flash Attention?FlashAttention fuer schnellere und speichersparende Attention. Wert on/off/auto je nach Build. | on |
| Kontext?Groesse des Kontextfensters in Token. 0 = der beim Training verwendete Kontext des Modells. | 26624 |
| Parallel?Anzahl paralleler Slots/Sequenzen, die der Server gleichzeitig bedient. Der Kontext wird auf die Slots aufgeteilt. | 12 |
Anzeige
Model comparison
OpenReasoning-Nemotron-32B 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 GeForce RTX 5090 481,0 tok/sNVIDIA RTX PRO 6000 Blackwell Workstation Edition 480,0 tok/sNVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition 411,7 tok/sNVIDIA GeForce RTX 3090 Ti 236,8 tok/sNVIDIA RTX A6000 144,0 tok/sAMD Radeon RX 7900 XTX 124,6 tok/sAMD Radeon AI PRO R9700 116,5 tok/sNVIDIA GeForce RTX 5070 Ti 49,9 tok/s★ AMD Radeon 8060S Graphics 16,5 tok/s this runNVIDIA GeForce RTX 2060 2,4 tok/sCPU-only 1,1 tok/s
CPUby processor
AMD Ryzen 7 5800X3D 8-Core Processor 481,0 tok/sAMD Ryzen 9 9950X 16-Core Processor 480,0 tok/sAMD Ryzen Threadripper PRO 9965WX 24-Cores 411,7 tok/sAMD Ryzen 9 8945HX with Radeon Graphics 236,8 tok/sAMD Ryzen Threadripper PRO 7955WX 16-Cores 144,0 tok/sIntel(R) Xeon(R) CPU E5-4657L v2 @ 2.40GHz 124,6 tok/sAMD Ryzen 3 3100 4-Core Processor 116,5 tok/sAMD Ryzen Threadripper PRO 5975WX 32-Cores 49,9 tok/s★ AMD RYZEN AI MAX+ 395 w/ Radeon 8060S 16,5 tok/s this runIntel(R) Xeon(R) CPU E5-4620 v2 @ 2.60GHz 1,1 tok/s
MBby mainboard
ASUSTeK COMPUTER INC. ROG STRIX B550-A GAMING 481,0 tok/sASUSTeK COMPUTER INC. ProArt X870E-CREATOR WIFI 480,0 tok/sASUSTeK COMPUTER INC. Pro WS WRX90E-SAGE SE 411,7 tok/sMeigao Innovation Technology (Shen Zhen) Co., Ltd DRFXL (MotherBoard Series) 236,8 tok/sDell Inc. PowerEdge R820 124,6 tok/sASUSTeK COMPUTER INC. ROG STRIX X570-F GAMING 116,5 tok/sASUSTeK COMPUTER INC. Pro WS WRX80E-SAGE SE WIFI 49,9 tok/s★ Meigao Innovation Technology (Shen Zhen) Co., Ltd SHWSA (MS-S1 MAX) 16,5 tok/s this run
DRVby driver
unbekannt 481,0 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
18 W
⚡ TDP
120 W
☁️ External LLM (API)
Electricity0.30 EUR/kWh
Avg power (incl. idle)120 W estimated (TDP)GPU 120 W full load
Avg cost / hourEUR 0.036
Electricity / 1M tokensEUR 0.60
Token / kWh495.90K
Acquisition (system)EUR 7,454 partial priceGPU EUR 3,000 · Board EUR 3,900 · RAM EUR 434 · PSU EUR 120
Electricity (2 years)–
TCO (2 years)EUR 8,085
Output tokens (2 years)1.04B
☁️ 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 (18 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
| OpenReasoning-Nemotron-32BAMD Radeon 8060S Graphics | OpenReasoning-Nemotron-32BNVIDIA GeForce RTX 5090 | OpenReasoning-Nemotron-32BNVIDIA RTX PRO 6000 Blackwell Workstation Edition | OpenReasoning-Nemotron-32B3x 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) | – | – | – | – |
AMD Radeon 8060S Graphics · this run 17 tok/sNVIDIA GeForce RTX 5090 481 tok/sNVIDIA RTX PRO 6000 Blackwell Workstation Edition 480 tok/s3× NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition 412 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.