ℹ

Public note: NVIDIA GB10 (DGX Spark, 120GB Unified Memory) . gpt-oss-120b Q4_K_M (2 Shards, 58,5 GiB) . llama.cpp b10507 native aarch64, vollstaendig im Unified Memory (kein Offload, kein Paging), -ngl 999, -np 12, -c 49152

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.

Generation135,08tok/s

Prefill1.547,86tok/s

Time to First Token12.948,00ms

Total duration180,55s

Concurrency10parallel

Ranking in the field

This GPU All GPUs

1048of 1479 systems

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

This run is better than 29 % of all comparable systems.

Generation 135,1 tok/s

-70 % vs Ø 445,0

Prefill 1.547,9 tok/s

-77 % vs Ø 6.823,3

Time to First Token 12.948 ms

-71 % vs Ø 44.468

Distribution in the field0 – 2.491 tok/s

Ø 445 Median Dieser Lauf

Wie schlägt sich dieser Benchmark mit anderen Modellen?

gemma-4-E2B-itNVIDIA GeForce RTX 5090 · run-20260728-184455-e28775

2.491,2 tok/s

gemma-4-E2B-itNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260727-181606-b37ed9

2.414,4 tok/s

gemma-4-E2B-itNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20261023-065752-48370d

2.412,3 tok/s

Nemotron-3.5-Lightning-30B-A3BNVIDIA GeForce RTX 3090 Ti · run-20261021-093815-dce558

2.263,7 tok/s

Nemotron-3-Nano-4BNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260727-181607-4cca42

2.182,7 tok/s

Nemotron-3-Nano-4BNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20261023-065758-96fe6a

2.144,3 tok/s

gemma-4-E2B-it3× 3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260728-184456-d87c73

2.143,6 tok/s

gpt-oss-20bNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260730-035052-9c966e

1.995,4 tok/s

gpt-oss-20bNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260728-140954-8007c4

1.993,5 tok/s

gpt-oss-20bNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20261023-065808-332d1f

1.981,1 tok/s

Nemotron-3-Nano-4B3× 3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260728-194136-0aed86

1.937,7 tok/s

Nemotron-3-Nano-4B3× 3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260728-194136-8db0fa

1.911,6 tok/s

Nemotron-3-Nano-4B3× 3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260727-181606-e0eafd

1.897,0 tok/s

gpt-oss-20bNVIDIA GeForce RTX 5090 · run-20260729-032121-7c961d

1.890,7 tok/s

gpt-oss-120b this runNVIDIA DGX SPARK (GB10) · run-20261024-122736-e9fae0

135,1 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-20260728-140953-f26a76

1.335,9 tok/s

NVIDIA RTX PRO 6000 Blackwell Workstation Editionrun-20260729-224611-44e81d

1.274,7 tok/s

3× NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Editionrun-20260728-034051-d52467

1.229,7 tok/s

3× NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Editionrun-20260729-032120-5069d9

1.220,8 tok/s

3× NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Editionrun-20260729-032120-8c69d5

1.214,9 tok/s

3× NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Editionrun-20260729-032120-67f4b8

1.122,3 tok/s

2× AMD Radeon PRO W7900 Dual Slotrun-20260818-172656-ec2b33

209,8 tok/s

NVIDIA DGX SPARK (GB10) this runrun-20261024-122736-e9fae0

135,1 tok/s

NVIDIA RTX SPARK (N1X)run-20261024-143953-835e9c

111,9 tok/s

3× AMD Radeon AI PRO R9700run-20260806-094434-540814

107,0 tok/s

3× AMD Radeon AI PRO R9700run-20260806-094435-c5a756

106,9 tok/s

3× AMD Radeon AI PRO R9700run-20260806-094435-9251c8

105,4 tok/s

NVIDIA GeForce RTX 3090 Tirun-20260728-034052-2683e4

85,3 tok/s

NVIDIA GeForce RTX 5070 Tirun-20260727-145956-fb4fba

60,1 tok/s

Hardware

GPU: NVIDIA DGX SPARK (GB10) · 128 GB VRAM
CPU: NVIDIA Grace
RAM: 120 GB
Mainboard: ASUSTeK COMPUTER INC. GX10

Setup

Runtime: llama.cpp
Quantization: Q4_K_M
Driver: NVIDIA 590.48.01 / CUDA 13.1
Model: gpt-oss-120b

Configuration

benchmark-konfiguration — run-20261024-122736-e9fae0

# LLM-Benchmark Konfiguration # Modell : gpt-oss-120b # Engine : llama.cpp # Run-ID : run-20261024-122736-e9fae0 # GPU : NVIDIA DGX SPARK (GB10) # CPU : NVIDIA Grace # RAM : 120 GB bench@llm-benchmark:~$ llama-server \ -m /home/godcore/bench_models/openai_gpt-oss-120b-Q4_K_M-00001-of-00002.gguf \ -ngl 999 \ -np 12 \ -c 49152

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.	gpt-oss-120b
Kontextlaenge?Maximale Anzahl Tokens (Eingabe + erzeugte Ausgabe zusammen), die das Modell pro Anfrage verarbeiten kann.	49152
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
np	12
Kontext?Groesse des Kontextfensters in Token. 0 = der beim Training verwendete Kontext des Modells.	49152
execution_type	local

All benchmarks of this model To leaderboard

Model comparison

gpt-oss-120b 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 1.335,9 tok/sNVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition 1.229,7 tok/sAMD Radeon PRO W7900 Dual Slot 216,2 tok/s★ NVIDIA DGX SPARK (GB10) 135,1 tok/s this runNVIDIA RTX SPARK (N1X) 111,9 tok/sAMD Radeon AI PRO R9700 107,0 tok/sNVIDIA GeForce RTX 3090 Ti 85,3 tok/sNVIDIA GeForce RTX 5070 Ti 60,1 tok/sNVIDIA GeForce RTX 5090 42,3 tok/sIntel Arc Pro B70 32,4 tok/sNVIDIA GeForce RTX 3090 18,1 tok/sAMD Radeon RX 7900 XTX 16,3 tok/sNVIDIA GeForce RTX 2060 6,8 tok/sCPU-only 5,3 tok/s

CPUby processor

AMD Ryzen 9 9950X 16-Core Processor 1.335,9 tok/sAMD Ryzen Threadripper PRO 9965WX 24-Cores 1.229,7 tok/sAMD Ryzen Threadripper PRO 5975WX 32-Cores 216,2 tok/s★ NVIDIA Grace 135,1 tok/s this runNVIDIA RTX Spark N1X (6144-core GPU, 20-core CPU) 111,9 tok/sAMD Ryzen Threadripper PRO 7955WX 16-Cores 107,0 tok/sAMD Ryzen 9 8945HX with Radeon Graphics 85,3 tok/sAMD Ryzen 7 5800X3D 8-Core Processor 42,3 tok/sAMD Ryzen 9 7945HX with Radeon Graphics 32,4 tok/sAMD Ryzen 5 5600X 6-Core Processor 18,1 tok/sIntel(R) Xeon(R) CPU E5-4657L v2 @ 2.40GHz 16,3 tok/sIntel(R) Xeon(R) CPU E5-4620 v2 @ 2.60GHz 5,3 tok/s

MBby mainboard

ASUSTeK COMPUTER INC. ProArt X870E-CREATOR WIFI 1.335,9 tok/sASUSTeK COMPUTER INC. Pro WS WRX90E-SAGE SE 1.229,7 tok/sASUSTeK COMPUTER INC. Pro WS WRX80E-SAGE SE WIFI 216,2 tok/s★ ASUSTeK COMPUTER INC. GX10 135,1 tok/s this runASUS H7607BA 111,9 tok/sMeigao Innovation Technology (Shen Zhen) Co., Ltd DRFXL (MotherBoard Series) 85,3 tok/sASUSTeK COMPUTER INC. ROG STRIX B550-A GAMING 42,3 tok/sShenzhen Meigao Electronic Equipment Co.,Ltd DRFXI (MotherBoard Series) 32,4 tok/sDell Inc. PowerEdge R820 16,3 tok/s

ENGby engine

★ llama.cpp 1.335,9 tok/s this rununbekannt 99,8 tok/svLLM 21,5 tok/s

DRVby driver

unbekannt 1.335,9 tok/s★ NVIDIA 590.48.01 / CUDA 13.1 135,1 tok/s this runAMD 7.0.0-27-generic 99,8 tok/sIntel 26.18.38308.4 32,4 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 URL Reset

⚡ Electricity price EUR/kWh

⚙️ System utilization 100 %

🖥️ Acquisition EUR

🔌 Idle 15 W

⚡ TDP 140 W

☁️ External LLM (API) show

Input €/1M Output €/1M Input per output token

Electricity0.30 EUR/kWh

Avg power (incl. idle)140 W estimated (TDP)GPU 140 W full load

Avg cost / hourEUR 0.042

Electricity / 1M tokensEUR 0.086

Token / kWh3.47M

Acquisition (system)EUR 5,800 partial priceGPU EUR 4,000 · RAM EUR 1,680 · PSU EUR 120

Electricity (2 years)–

TCO (2 years)EUR 6,536

Output tokens (2 years)8.52B

☁️ 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 (15 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

	gpt-oss-120bNVIDIA DGX SPARK (GB10)	gpt-oss-120bNVIDIA RTX PRO 6000 Blackwell Workstation Edition	gpt-oss-120bNVIDIA RTX PRO 6000 Blackwell Workstation Edition	gpt-oss-120b3x 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)	–	–	–	–

NVIDIA DGX SPARK (GB10) · this run 135 tok/s NVIDIA RTX PRO 6000 Blackwell Workstation Edition 1,336 tok/s NVIDIA RTX PRO 6000 Blackwell Workstation Edition 1,275 tok/s 3× NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition 1,230 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.