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

Nemotron-H-47B-Reasoning-128K

Performance benchmark · measured on 29.07.2026 02:22

Benchmark-IDrun-20260729-032131-d57c05
Timebench 3 - Kombi (Prefill + Generation)Dense47BRuntime: llama.cppQuantisierung: Q4_K_M
Generation174,90tok/s
Prefill848,29tok/s
Time to First Token34.870,00ms
Total duration88,99s
Concurrency10parallel
Ranking in the field
57of 64 systems

Performance benchmark · Primary metric: Generation-Speed (tok/s) · 10× concurrent

This run is better than 11 % of all comparable systems.
Generation 174,9 tok/s
-82 % vs Ø 997,9
Prefill 848,3 tok/s
-92 % vs Ø 10.841,0
Time to First Token 34.870 ms
+41 % vs Ø 24.699
Distribution in the field0 – 2.414 tok/s
Ø 998 Median Dieser Lauf

Wie schlägt sich dieser Benchmark mit anderen Modellen?

gemma-4-E2B-itNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260727-181606-b37ed9
2.414,4 tok/s
NVIDIA-Nemotron-3-Nano-4BNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260727-181607-4cca42
2.182,7 tok/s
gpt-oss-20bNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260728-140954-8007c4
1.993,5 tok/s
Nemotron-3-Nano-Omni-30B-A3B-ReasoningNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260729-032119-311459
1.791,5 tok/s
NVIDIA-Nemotron-3-Nano-30B-A3BNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260729-032121-e2403f
1.773,5 tok/s
Nemotron-Cascade-2-30B-A3BNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260729-032120-1b1a49
1.768,1 tok/s
gemma-4-E4B-itNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260727-181607-866815
1.758,6 tok/s
Laguna-XS-2.1NVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260728-140955-9f055e
1.757,4 tok/s
Laguna-XS-2.1NVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260728-140955-f34b37
1.753,8 tok/s
Laguna-XS-2.1NVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260728-140955-1e8c74
1.740,4 tok/s
Mamba-Codestral-7B-v0.1NVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260728-194136-5b0cb8
1.712,6 tok/s
Qwen3-Coder-30B-A3B-InstructNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260729-032124-ef2bb1
1.690,0 tok/s
Qwen3-30B-A3B-Thinking-2507NVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260729-032123-88fa84
1.670,5 tok/s
Qwen3-30B-A3B-Instruct-2507NVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260729-032122-3d6d17
1.658,1 tok/s
Nemotron-H-47B-Reasoning-128K this runNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260729-032131-d57c05
174,9 tok/s

Configuration

benchmark-konfiguration — run-20260729-032131-d57c05
# LLM-Benchmark Konfiguration # Modell : Nemotron-H-47B-Reasoning-128K # Engine : llama.cpp # Run-ID : run-20260729-032131-d57c05 # GPU : NVIDIA RTX PRO 6000 Blackwell Workstation Edition # CPU : AMD Ryzen 9 9950X 16-Core Processor # RAM : 92 GB bench@llm-benchmark:~$ /home/godcore/llama.cpp/build/bin/llama-server \ -m /home/godcore/.cache/huggingface/hub/models--bartowski--nvidia_Nemotron-H-47B-Reasoning-128K-GGUF/snapshots/ce89f02cd0eafe3931acec38f4d91a654ae1ca4d/nvidia_Nemotron-H-47B-Reasoning-128K-Q4_K_M.gguf \ --alias Nemotron-H-47B-Reasoning-128K \ --host 0.0.0.0 \ --port 8000 \ -ngl 999 \ -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.Nemotron-H-47B-Reasoning-128K
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--bartowski--nvidia_Nemotron-H-47B-Reasoning-128K-GGUF/snapshots/ce89f02cd0eafe3931acec38f4d91a654ae1ca4d/nvidia_Nemotron-H-47B-Reasoning-128K-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.999
Kontext?Groesse des Kontextfensters in Token. 0 = der beim Training verwendete Kontext des Modells.16384
np4

All benchmarks of this model To leaderboard

Anzeige
Model comparison

Nemotron-H-47B-Reasoning-128K 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

22216711155,60,03215748271.079Prefill (tok/s)Generation (tok/s)AMD Radeon AI PRO R9700 - 24,9 tok/s Generation, 462 tok/s Prefill, TTFT 21.185 ms (1 Lauf)AMD Radeon AI PRO R97...NVIDIA RTX PRO 6000 Blackwell Workstation Edition - 174,9 tok/s Generation, 938 tok/s Prefill, TTFT 18.279 ms (3 Laufe) | DIESER LAUF★ NVIDIA RTX PRO 6000 B...
★ NVIDIA RTX PRO 6000 Blackwell Workstation Edition 174,9 tok/s this runAMD Radeon AI PRO R9700 24,9 tok/s

CPUby processor

22216711155,60,03215748271.079Prefill (tok/s)Generation (tok/s)AMD Ryzen Threadripper PRO 7955WX 16-Cores - 24,9 tok/s Generation, 462 tok/s Prefill, TTFT 21.185 ms (1 Lauf)AMD Ryzen Threadrippe...AMD Ryzen 9 9950X 16-Core Processor - 174,9 tok/s Generation, 938 tok/s Prefill, TTFT 18.279 ms (3 Laufe) | DIESER LAUF★ AMD Ryzen 9 9950X 16-...
★ AMD Ryzen 9 9950X 16-Core Processor 174,9 tok/s this runAMD Ryzen Threadripper PRO 7955WX 16-Cores 24,9 tok/s

MBby mainboard

22216711155,60,03215748271.079Prefill (tok/s)Generation (tok/s)ASUSTeK COMPUTER INC. Pro WS WRX90E-SAGE SE - 24,9 tok/s Generation, 462 tok/s Prefill, TTFT 21.185 ms (1 Lauf)ASUSTeK COMPUTER INC....ASUSTeK COMPUTER INC. ProArt X870E-CREATOR WIFI - 174,9 tok/s Generation, 938 tok/s Prefill, TTFT 18.279 ms (3 Laufe) | DIESER LAUF★ ASUSTeK COMPUTER INC....
★ ASUSTeK COMPUTER INC. ProArt X870E-CREATOR WIFI 174,9 tok/s this runASUSTeK COMPUTER INC. Pro WS WRX90E-SAGE SE 24,9 tok/s

ENGby engine

192184175166157770802835868Prefill (tok/s)Generation (tok/s)llama.cpp - 174,9 tok/s Generation, 819 tok/s Prefill, TTFT 19.005 ms (4 Laufe) | DIESER LAUF★ llama.cpp
★ llama.cpp 174,9 tok/s this run

DRVby driver

192184175166157770802835868Prefill (tok/s)Generation (tok/s)unbekannt - 174,9 tok/s Generation, 819 tok/s Prefill, TTFT 19.005 ms (4 Laufe)unbekannt
unbekannt 174,9 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 70 W
⚡ TDP 644 W
☁️ External LLM (API)
Electricity0.30 EUR/kWh
Avg power (incl. idle)644 W estimated (TDP)GPU 600 + CPU 29 + Board 15 W full load
Avg cost / hourEUR 0.19
Electricity / 1M tokensEUR 0.31
Token / kWh978.08K
Acquisition (system)EUR 15,248 full priceGPU EUR 13,000 · CPU EUR 649 · Board EUR 499 · RAM EUR 920 · PSU EUR 180
Electricity (2 years)
TCO (2 years)EUR 18,632
Output tokens (2 years)11.03B
☁️ 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 (70 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

Nemotron-H-47B-Reasoning-128KNVIDIA 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)

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.