Contributed byMario AlkaNVIDIA NVIDIA-Nemotron-3-Nano-30B-A3B
Performance benchmark · measured on 29.07.2026 01:06
Benchmark-ID
run-20260729-032126-267861Timebench 3 - Kombi (Prefill + Generation)MoE30BRuntime: llama.cppQuantisierung: Q4_K_M
Generation263,34tok/s
Prefill784,42tok/s
Time to First Token55.800,00ms
Total duration281,58s
Concurrency10parallel
Ranking in the field
22of 76 systems
Performance benchmark · Primary metric: Generation-Speed (tok/s) · 10× concurrent
This run is better than 72 % of all comparable systems.
Generation
263,3 tok/s
-5 % vs Ø 275,8
Prefill
784,4 tok/s
-68 % vs Ø 2.440,1
Time to First Token
55.800 ms
-45 % vs Ø 101.921
Distribution in the field2 – 1.744 tok/s
Wie schlägt sich dieser Benchmark mit anderen Modellen?
gemma-4-E2B-itNVIDIA GeForce RTX 5070 Ti · run-20260728-140953-875f2e
1.744,0 tok/s
gemma-4-E2B-itNVIDIA GeForce RTX 5070 Ti · run-20260727-090152-d5613d
1.743,1 tok/s
gemma-4-E2B-itNVIDIA GeForce RTX 5070 Ti · run-20260727-032758-bb6a92
1.608,5 tok/s
gpt-oss-20bNVIDIA GeForce RTX 5070 Ti · run-20260727-145956-ba1dd9
1.210,3 tok/s
NVIDIA-Nemotron-3-Nano-4BNVIDIA GeForce RTX 5070 Ti · run-20260727-090153-f232e7
1.197,0 tok/s
gemma-4-E4B-itNVIDIA GeForce RTX 5070 Ti · run-20260727-090153-f5c391
1.082,4 tok/s
gemma-4-E4B-itNVIDIA GeForce RTX 5070 Ti · run-20260727-032759-95c93c
1.022,8 tok/s
Gemma-4-26B-A4B-itNVIDIA GeForce RTX 5070 Ti · run-20260729-032121-0de27f
1.009,0 tok/s
Meta-Llama-3.1-8B-InstructNVIDIA GeForce RTX 5070 Ti · run-20260728-194135-ea27af
913,4 tok/s
Ornith-1.0-9BNVIDIA GeForce RTX 5070 Ti · run-20260730-064527-e00abe
883,4 tok/s
gemma-4-12B-itNVIDIA GeForce RTX 5070 Ti · run-20260728-194136-c688b4
677,5 tok/s
Ministral-3-14B-Reasoning-2512NVIDIA GeForce RTX 5070 Ti · run-20260728-194137-c91fd6
590,0 tok/s
Ministral-3-14B-Reasoning-2512NVIDIA GeForce RTX 5070 Ti · run-20260727-032759-f155dc
497,6 tok/s
gemma-4-12B-itNVIDIA GeForce RTX 5070 Ti · run-20260727-032759-8c9d3d
486,2 tok/s
NVIDIA-Nemotron-3-Nano-30B-A3B this runNVIDIA GeForce RTX 5070 Ti · run-20260729-032126-267861
263,3 tok/s
How does this benchmark compare on other GPUs?
Same model on different hardware · 10× concurrent · Generation (tok/s)
NVIDIA RTX PRO 6000 Blackwell Workstation Editionrun-20261023-065821-3781a7
1.775,6 tok/s
NVIDIA RTX PRO 6000 Blackwell Workstation Editionrun-20260729-032121-e2403f
1.773,5 tok/s
NVIDIA GeForce RTX 5090run-20260729-105335-23842b
1.722,1 tok/s
3× NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Editionrun-20260729-105341-be087a
1.654,6 tok/s
3× NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Editionrun-20260729-105341-bdfbe0
1.554,4 tok/s
3× NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Editionrun-20260729-105340-283a81
1.540,5 tok/s
NVIDIA GeForce RTX 3090 Tirun-20260729-060802-d2028a
897,5 tok/s
2× NVIDIA RTX A6000run-20260827-221527-db993e
756,7 tok/s
2× NVIDIA RTX A6000run-20260830-051029-eecc59
755,1 tok/s
NVIDIA RTX A6000run-20260826-211812-648be8
588,5 tok/s
NVIDIA RTX PRO 6000 Blackwell Workstation Editionrun-20260724-082937-032cbf
533,5 tok/s
NVIDIA GeForce RTX 3090 Tirun-20260804-052131-349ba0
404,7 tok/s
2× NVIDIA RTX A6000run-20260829-212822-e2f28e
403,7 tok/s
2× NVIDIA RTX A6000run-20260827-154104-068025
403,6 tok/s
Hardware
GPU: NVIDIA GeForce RTX 5070 Ti · 16 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: Q4_K_M
Model: NVIDIA-Nemotron-3-Nano-30B-A3B
Configuration
# LLM-Benchmark Konfiguration
# Modell : NVIDIA-Nemotron-3-Nano-30B-A3B
# Engine : llama.cpp
# Run-ID : run-20260729-032126-267861
# GPU : NVIDIA GeForce RTX 5070 Ti
# CPU : AMD Ryzen Threadripper PRO 5975WX 32-Cores
# RAM : 247 GB
bench@llm-benchmark:~$ /home/godcore/llama.cpp/build/bin/llama-server \
-m /home/godcore/.cache/huggingface/hub/models--DevQuasar--nvidia.NVIDIA-Nemotron-3-Nano-30B-A3B-BF16-GGUF/snapshots/3730de2284499e33be97bb6e63efe1ec183abba2/nvidia.NVIDIA-Nemotron-3-Nano-30B-A3B-BF16.Q4_K_M.gguf \
--alias NVIDIA-Nemotron-3-Nano-30B-A3B \
--host 0.0.0.0 \
--port 8000 \
-ngl 30 \
-c 16384 \
-np 4 \
--jinja
| 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. | NVIDIA-Nemotron-3-Nano-30B-A3B |
| Kontextlaenge?Maximale Anzahl Tokens (Eingabe + erzeugte Ausgabe zusammen), die das Modell pro Anfrage verarbeiten kann. | 16384 |
| Modellpfad?Pfad zur GGUF-Modelldatei, die geladen und ausgeliefert wird. | /home/godcore/.cache/huggingface/hub/models--DevQuasar--nvidia.NVIDIA-Nemotron-3-Nano-30B-A3B-BF16-GGUF/snapshots/3730de2284499e33be97bb6e63efe1ec183abba2/nvidia.NVIDIA-Nemotron-3-Nano-30B-A3B-BF16.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. | 30 |
| Kontext?Groesse des Kontextfensters in Token. 0 = der beim Training verwendete Kontext des Modells. | 16384 |
| np | 4 |
| execution_type | local |
Anzeige
Model comparison
NVIDIA-Nemotron-3-Nano-30B-A3B 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 1.775,6 tok/sNVIDIA GeForce RTX 5090 1.722,1 tok/sNVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition 1.654,6 tok/sNVIDIA GeForce RTX 3090 Ti 897,5 tok/sNVIDIA RTX A6000 756,7 tok/sAMD Radeon AI PRO R9700 294,4 tok/s★ NVIDIA GeForce RTX 5070 Ti 263,3 tok/s this runAMD Radeon RX 7900 XTX 244,9 tok/sNVIDIA DGX SPARK (GB10) 58,3 tok/sNVIDIA GeForce RTX 2060 15,1 tok/sCPU-only 6,6 tok/s
CPUby processor
AMD Ryzen 5 5600X 6-Core Processor 1.775,6 tok/sAMD Ryzen 9 9950X 16-Core Processor 1.773,5 tok/sAMD Ryzen 7 5800X3D 8-Core Processor 1.722,1 tok/sAMD Ryzen Threadripper PRO 9965WX 24-Cores 1.654,6 tok/sAMD Ryzen 9 8945HX with Radeon Graphics 897,5 tok/sAMD Ryzen Threadripper PRO 7955WX 16-Cores 756,7 tok/sAMD Ryzen 3 3100 4-Core Processor 294,4 tok/s★ AMD Ryzen Threadripper PRO 5975WX 32-Cores 263,3 tok/s this runIntel(R) Xeon(R) CPU E5-4657L v2 @ 2.40GHz 244,9 tok/sNVIDIA Grace 58,3 tok/sIntel(R) Xeon(R) CPU E5-4620 v2 @ 2.60GHz 6,6 tok/s
MBby mainboard
ASUSTeK COMPUTER INC. ROG STRIX B550-A GAMING 1.775,6 tok/sASUSTeK COMPUTER INC. ProArt X870E-CREATOR WIFI 1.773,5 tok/sASUSTeK COMPUTER INC. Pro WS WRX90E-SAGE SE 1.654,6 tok/sMeigao Innovation Technology (Shen Zhen) Co., Ltd DRFXL (MotherBoard Series) 897,5 tok/sASUSTeK COMPUTER INC. PRIME A520M-K 404,7 tok/sASUSTeK COMPUTER INC. ROG STRIX X570-F GAMING 294,4 tok/s★ ASUSTeK COMPUTER INC. Pro WS WRX80E-SAGE SE WIFI 263,3 tok/s this runDell Inc. PowerEdge R820 244,9 tok/sASUSTeK COMPUTER INC. GX10 58,3 tok/s
DRVby driver
unbekannt 1.775,6 tok/sNVIDIA 590.48.01 / CUDA 13.1 58,3 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
310 W
☁️ External LLM (API)
Electricity0.30 EUR/kWh
Avg power (incl. idle)310 W estimated (TDP)GPU 300 + Board 10 W full load
Avg cost / hourEUR 0.093
Electricity / 1M tokensEUR 0.098
Token / kWh3.06M
Acquisition (system)EUR 5,000 partial priceGPU EUR 994 · CPU EUR 1,880 · RAM EUR 1,976 · PSU EUR 150
Electricity (2 years)–
TCO (2 years)EUR 6,629
Output tokens (2 years)16.61B
☁️ 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
| NVIDIA-Nemotron-3-Nano-30B-A3BNVIDIA GeForce RTX 5070 Ti | NVIDIA-Nemotron-3-Nano-30B-A3BNVIDIA RTX PRO 6000 Blackwell Workstation Edition | NVIDIA-Nemotron-3-Nano-30B-A3BNVIDIA RTX PRO 6000 Blackwell Workstation Edition | NVIDIA-Nemotron-3-Nano-30B-A3BNVIDIA GeForce RTX 5090 | |
|---|---|---|---|---|
| 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) | – | – | – | – |
NVIDIA GeForce RTX 5070 Ti · this run 263 tok/sNVIDIA RTX PRO 6000 Blackwell Workstation Edition 1,776 tok/sNVIDIA RTX PRO 6000 Blackwell Workstation Edition 1,774 tok/sNVIDIA GeForce RTX 5090 1,722 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.