Created bymario-alka.dePowered bygodcore.denoob2claw.detricoma.de
Contributed byMario AlkaNVIDIA

NVIDIA-Nemotron-3-Nano-30B-A3B

Performance benchmark · measured on 23.07.2026 00:27

Benchmark-IDrun-20260728-195846-3e95b4
Timebench 3 - Kombi (Prefill + Generation)MoE30BRuntime: llama.cppQuantisierung: Q4_K_M
Generation6,60tok/s
Prefill73,44tok/s
Time to First Token150.848,50ms
Total duration600,00s
Concurrency5parallel
Ranking in the field
4of 11 systems

Performance benchmark · Primary metric: Generation-Speed (tok/s) · 5× concurrent

This run is better than 70 % of all comparable systems.
Generation 6,6 tok/s
+10 % vs Ø 6,0
Prefill 73,4 tok/s
+6 % vs Ø 69,1
Time to First Token 150.849 ms
-7 % vs Ø 163.034
Distribution in the field0 – 10 tok/s
Ø 6 Median Dieser Lauf

Wie schlägt sich dieser Benchmark mit anderen Modellen?

How does this benchmark compare on other GPUs?

Same model on different hardware · 5× concurrent · Generation (tok/s)

Hardware

GPU: Keine GPU (CPU-only)
CPU: 4x Intel(R) Xeon(R) CPU E5-4620 v2 @ 2.60GHz
RAM: 504 GB
Mainboard: Dell Inc. PowerEdge R820

Setup

Runtime: llama.cpp
Quantization: Q4_K_M
Model: NVIDIA-Nemotron-3-Nano-30B-A3B

Configuration

benchmark-konfiguration — run-20260728-195846-3e95b4
# LLM-Benchmark Konfiguration # Modell : NVIDIA-Nemotron-3-Nano-30B-A3B # Engine : llama.cpp # Run-ID : run-20260728-195846-3e95b4 # GPU : Keine GPU (CPU-only) # CPU : 4x Intel(R) Xeon(R) CPU E5-4620 v2 @ 2.60GHz # RAM : 504 GB bench@llm-benchmark:~$ /opt/llama.cpp/build/bin/llama-server \ -m /opt/models/Nemotron-3-Nano-30B-A3B-Q4_K_M.gguf \ -a Nemotron-3-Nano-30B-A3B \ --host 0.0.0.0 \ --port 8080 \ --numa distribute \ -t 64 \ -tb 64 \ -c 8192 \ --parallel 4
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.Nemotron-3-Nano-30B-A3B
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./opt/models/Nemotron-3-Nano-30B-A3B-Q4_K_M.gguf
Alias?Anzeigename des Modells nach aussen (served model name), unabhaengig vom Dateinamen.Nemotron-3-Nano-30B-A3B
NUMA?NUMA-Optimierung fuer Multi-Socket-CPUs: distribute/isolate/numactl. Verbessert die Speicherlokalitaet.distribute
Threads?Anzahl CPU-Threads fuer die Token-Generierung (Decode).64
Batch-Threads?Anzahl CPU-Threads fuer Prompt-Verarbeitung und Batch (Prefill).64
Kontext?Groesse des Kontextfensters in Token. 0 = der beim Training verwendete Kontext des Modells.8192
Parallel?Anzahl paralleler Slots/Sequenzen, die der Server gleichzeitig bedient. Der Kontext wird auf die Slots aufgeteilt.4

All benchmarks of this model To leaderboard

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

6925193461730,004.6959.39014.086Prefill (tok/s)Generation (tok/s)NVIDIA RTX PRO 6000 Blackwell Workstation Edition - 533,5 tok/s Generation, 11.370 tok/s Prefill, TTFT 2.636 ms (3 Laufe)NVIDIA RTX PRO 6000 B...AMD Radeon AI PRO R9700 - 274,0 tok/s Generation, 10.797 tok/s Prefill, TTFT 2.446 ms (4 Laufe)AMD Radeon AI PRO R97...AMD Radeon 8060S Graphics - 118,7 tok/s Generation, 1.158 tok/s Prefill, TTFT 9.116 ms (6 Laufe)AMD Radeon 8060S Grap...CPU-only - 6,6 tok/s Generation, 72 tok/s Prefill, TTFT 131.051 ms (3 Laufe) | DIESER LAUF★ CPU-only
NVIDIA RTX PRO 6000 Blackwell Workstation Edition 533,5 tok/sAMD Radeon AI PRO R9700 274,0 tok/sAMD Radeon 8060S Graphics 118,7 tok/s★ CPU-only 6,6 tok/s this run

CPUby processor

6925193461730,004.6959.39014.086Prefill (tok/s)Generation (tok/s)AMD Ryzen 9 9950X 16-Core Processor - 533,5 tok/s Generation, 11.370 tok/s Prefill, TTFT 2.636 ms (3 Laufe)AMD Ryzen 9 9950X 16-...AMD Ryzen Threadripper PRO 7955WX 16-Cores - 274,0 tok/s Generation, 10.797 tok/s Prefill, TTFT 2.446 ms (4 Laufe)AMD Ryzen Threadrippe...AMD RYZEN AI MAX+ 395 w/ Radeon 8060S - 118,7 tok/s Generation, 1.158 tok/s Prefill, TTFT 9.116 ms (6 Laufe)AMD RYZEN AI MAX+ 395...Intel(R) Xeon(R) CPU E5-4620 v2 @ 2.60GHz - 6,6 tok/s Generation, 72 tok/s Prefill, TTFT 131.051 ms (3 Laufe) | DIESER LAUF★ Intel(R) Xeon(R) CPU ...
AMD Ryzen 9 9950X 16-Core Processor 533,5 tok/sAMD Ryzen Threadripper PRO 7955WX 16-Cores 274,0 tok/sAMD RYZEN AI MAX+ 395 w/ Radeon 8060S 118,7 tok/s★ Intel(R) Xeon(R) CPU E5-4620 v2 @ 2.60GHz 6,6 tok/s this run

MBby mainboard

6925193461730,004.6959.39014.086Prefill (tok/s)Generation (tok/s)ASUSTeK COMPUTER INC. ProArt X870E-CREATOR WIFI - 533,5 tok/s Generation, 11.370 tok/s Prefill, TTFT 2.636 ms (3 Laufe)ASUSTeK COMPUTER INC....ASUSTeK COMPUTER INC. Pro WS WRX90E-SAGE SE - 274,0 tok/s Generation, 10.797 tok/s Prefill, TTFT 2.446 ms (4 Laufe)ASUSTeK COMPUTER INC....Bosgame AXB35-02 (BeyondMax Series) - 118,7 tok/s Generation, 1.158 tok/s Prefill, TTFT 9.116 ms (6 Laufe)Bosgame AXB35-02 (Bey...Dell Inc. PowerEdge R820 - 6,6 tok/s Generation, 72 tok/s Prefill, TTFT 131.051 ms (3 Laufe) | DIESER LAUF★ Dell Inc. PowerEdge R...
ASUSTeK COMPUTER INC. ProArt X870E-CREATOR WIFI 533,5 tok/sASUSTeK COMPUTER INC. Pro WS WRX90E-SAGE SE 274,0 tok/sBosgame AXB35-02 (BeyondMax Series) 118,7 tok/s★ Dell Inc. PowerEdge R820 6,6 tok/s this run

ENGby engine

63952140428616905.53811.07616.615Prefill (tok/s)Generation (tok/s)vLLM - 533,5 tok/s Generation, 13.615 tok/s Prefill, TTFT 2.206 ms (5 Laufe)vLLMllama.cpp - 274,0 tok/s Generation, 1.489 tok/s Prefill, TTFT 41.319 ms (11 Laufe) | DIESER LAUF★ llama.cpp
vLLM 533,5 tok/s★ llama.cpp 274,0 tok/s this run

DRVby driver

5875605345074804.9625.1735.3845.595Prefill (tok/s)Generation (tok/s)unbekannt - 533,5 tok/s Generation, 5.279 tok/s Prefill, TTFT 29.096 ms (16 Laufe)unbekannt
unbekannt 533,5 tok/s
💰 Economics

Economics of this run

Operating cost, TCO and comparison with the next-best runs of the same model at identical concurrency (5× 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 35 W
⚡ TDP 105 W
☁️ External LLM (API)
Electricity0.30 EUR/kWh
Avg power (incl. idle)105 W estimated (TDP)CPU 95 + Board 10 W full load
Avg cost / hourEUR 0.032
Electricity / 1M tokensEUR 1.33
Token / kWh226.29K
Acquisition (system)EUR 2,175 partial priceCPU EUR 39 · RAM EUR 2,016 · PSU EUR 120
Electricity (2 years)
TCO (2 years)EUR 2,727
Output tokens (2 years)416.28M
☁️ 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 (35 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-A3BKeine GPU (CPU-only)NVIDIA-Nemotron-3-Nano-30B-A3BNVIDIA RTX PRO 6000 Blackwell Workstation EditionNVIDIA-Nemotron-3-Nano-30B-A3B3x AMD Radeon AI PRO R9700NVIDIA-Nemotron-3-Nano-30B-A3BAMD Radeon 8060S Graphics
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.