ℹ

Public note: NVIDIA GB10 (DGX Spark, 120GB Unified Memory) . gpt-oss-20b Q4_K_M . llama.cpp b10507 native aarch64, -ngl 999, -np 12, -c 49152 (4096 Tok/Slot)

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.

Generation156,02tok/s

Prefill3.859,29tok/s

Time to First Token2.569,50ms

Total duration72,18s

Concurrency5parallel

Ranking in the field

This GPU All GPUs

805of 1510 systems

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

This run is better than 47 % of all comparable systems.

Generation 156,0 tok/s

-39 % vs Ø 256,1

Prefill 3.859,3 tok/s

-29 % vs Ø 5.434,2

Time to First Token 2.570 ms

-91 % vs Ø 29.420

Distribution in the field0 – 1.349 tok/s

Ø 256 Median Dieser Lauf

Wie schlägt sich dieser Benchmark mit anderen Modellen?

gemma-4-E2B-itNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260727-181606-0adf6d

1.349,3 tok/s

gemma-4-E2B-itNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20261023-071844-e5e309

1.348,4 tok/s

gemma-4-E2B-itNVIDIA GeForce RTX 5090 · run-20260728-184455-5b4937

1.301,9 tok/s

gemma-4-E2B-it3× 3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260728-184456-4111a9

1.195,7 tok/s

Nemotron-3-Nano-4BNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260727-181607-30d864

1.188,7 tok/s

Nemotron-3-Nano-4BNVIDIA GeForce RTX 5090 · run-20260728-194135-5432cd

1.164,8 tok/s

Nemotron-3-Nano-4BNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20261023-071851-c3c4a5

1.154,4 tok/s

gpt-oss-20b same modelNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260728-140954-9e81f9

1.136,3 tok/s

gpt-oss-20b same modelNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260730-035052-09627f

1.135,0 tok/s

Nemotron-3.5-Lightning-30B-A3BNVIDIA GeForce RTX 3090 Ti · run-20261021-093815-b22926

1.135,0 tok/s

gpt-oss-20b same modelNVIDIA GeForce RTX 5090 · run-20260729-032121-058a31

1.113,5 tok/s

gpt-oss-20b same modelNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20261023-071836-99dfa5

1.105,8 tok/s

Nemotron-3-Nano-4B3× 3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260728-194136-836a91

1.090,6 tok/s

Nemotron-3-Nano-4B3× 3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260727-181606-8b381e

1.077,4 tok/s

gpt-oss-20b this runNVIDIA GB10 (DGX Spark) · run-20261024-115637-9d9ed3

156,0 tok/s

How does this benchmark compare on other GPUs?

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

NVIDIA RTX PRO 6000 Blackwell Workstation Editionrun-20260728-140954-9e81f9

1.136,3 tok/s

NVIDIA RTX PRO 6000 Blackwell Workstation Editionrun-20260730-035052-09627f

1.135,0 tok/s

NVIDIA GeForce RTX 5090run-20260729-032121-058a31

1.113,5 tok/s

NVIDIA RTX PRO 6000 Blackwell Workstation Editionrun-20261023-071836-99dfa5

1.105,8 tok/s

3× NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Editionrun-20260729-032121-ffb18a

1.015,5 tok/s

3× NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Editionrun-20260728-034052-437388

989,7 tok/s

3× NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Editionrun-20260729-032120-8b6146

970,4 tok/s

3× NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Editionrun-20260729-032120-310458

946,5 tok/s

NVIDIA GeForce RTX 5090run-20260724-004608-4bbadc

848,3 tok/s

NVIDIA RTX PRO 6000 Blackwell Workstation Editionrun-20260723-194447-1679c5

842,9 tok/s

NVIDIA GeForce RTX 5090run-20260724-004604-32d132

804,0 tok/s

3× NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Editionrun-20260724-082940-819a98

790,1 tok/s

NVIDIA GeForce RTX 5090run-20260723-045447-5656a1

784,5 tok/s

NVIDIA GeForce RTX 5070 Tirun-20260727-145956-0a1a23

710,4 tok/s

Hardware

GPU: NVIDIA 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-20b

Configuration

benchmark-konfiguration — run-20261024-115637-9d9ed3

# LLM-Benchmark Konfiguration # Modell : gpt-oss-20b # Engine : llama.cpp # Run-ID : run-20261024-115637-9d9ed3 # GPU : NVIDIA GB10 # CPU : NVIDIA Grace # RAM : 120 GB bench@llm-benchmark:~$ llama-server \ -m /home/godcore/bench_models/openai_gpt-oss-20b-Q4_K_M.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-20b
Kontextlaenge?Maximale Anzahl Tokens (Eingabe + erzeugte Ausgabe zusammen), die das Modell pro Anfrage verarbeiten kann.	49152
Modellpfad?Pfad zur GGUF-Modelldatei, die geladen und ausgeliefert wird.	/home/godcore/bench_models/openai_gpt-oss-20b-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
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-20b 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 1.995,4 tok/sNVIDIA GeForce RTX 5090 1.890,7 tok/sNVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition 1.835,7 tok/sNVIDIA GeForce RTX 5070 Ti 1.210,3 tok/sNVIDIA GeForce RTX 3090 Ti 1.113,5 tok/sNVIDIA RTX A6000 703,8 tok/sNVIDIA GeForce RTX 3090 448,2 tok/sAMD Radeon RX 7900 XTX 357,7 tok/sAMD Radeon PRO W7900 Dual Slot 344,1 tok/sAMD Radeon AI PRO R9700 323,6 tok/sAMD Radeon PRO W7800 48GB 280,0 tok/s★ NVIDIA GB10 (DGX Spark) 233,3 tok/s this runNVIDIA Tesla V100 PCIe 32GB 206,6 tok/sNVIDIA RTX Spark N1X (6144-core Blackwell RTX GPU) 200,5 tok/sNVIDIA Tesla P100 PCIe 16GB 111,1 tok/sAMD Radeon 8060S Graphics 74,4 tok/sCPU-only 70,9 tok/s

CPUby processor

AMD Ryzen 9 9950X 16-Core Processor 1.995,4 tok/sAMD Ryzen 5 5600X 6-Core Processor 1.981,1 tok/sAMD Ryzen 7 5800X3D 8-Core Processor 1.890,7 tok/sAMD Ryzen Threadripper PRO 9965WX 24-Cores 1.835,7 tok/sAMD Ryzen Threadripper PRO 5975WX 32-Cores 1.210,3 tok/sAMD Ryzen 9 8945HX with Radeon Graphics 1.113,5 tok/sAMD Ryzen Threadripper PRO 7955WX 16-Cores 703,8 tok/sAMD Ryzen Threadripper PRO 3955WX 16-Cores 357,7 tok/sAMD Ryzen 3 3100 4-Core Processor 323,6 tok/sIntel(R) Xeon(R) CPU E5-4657L v2 @ 2.40GHz 247,3 tok/s★ NVIDIA Grace 233,3 tok/s this runAMD Ryzen 3 PRO 3200GE w/ Radeon Vega Graphics 206,6 tok/sNVIDIA RTX Spark N1X (6144-core GPU, 20-core CPU) 200,5 tok/sAMD Ryzen 9 7945HX with Radeon Graphics 111,1 tok/sAMD RYZEN AI MAX+ 395 w/ Radeon 8060S 74,4 tok/sIntel(R) Xeon(R) CPU E5-4620 v2 @ 2.60GHz 14,5 tok/s

MBby mainboard

ASUSTeK COMPUTER INC. ProArt X870E-CREATOR WIFI 1.995,4 tok/sASUSTeK COMPUTER INC. ROG STRIX B550-A GAMING 1.981,1 tok/sASUSTeK COMPUTER INC. Pro WS WRX90E-SAGE SE 1.835,7 tok/sASUSTeK COMPUTER INC. Pro WS WRX80E-SAGE SE WIFI 1.210,3 tok/sMeigao Innovation Technology (Shen Zhen) Co., Ltd DRFXL (MotherBoard Series) 1.113,5 tok/sASUSTeK COMPUTER INC. PRIME A520M-K 572,1 tok/sASUSTeK COMPUTER INC. ROG STRIX X570-F GAMING 323,6 tok/sDell Inc. PowerEdge R820 247,3 tok/s★ ASUSTeK COMPUTER INC. GX10 233,3 tok/s this runASUS H7607BA 200,5 tok/sShenzhen Meigao Electronic Equipment Co.,Ltd F1FXM (DeskMini Series) 111,1 tok/sMeigao Innovation Technology (Shen Zhen) Co., Ltd SHWSA (MS-S1 MAX) 74,4 tok/s

ENGby engine

★ llama.cpp 1.995,4 tok/s this runvLLM 1.443,9 tok/sunbekannt 147,8 tok/s

DRVby driver

unbekannt 1.995,4 tok/s★ NVIDIA 590.48.01 / CUDA 13.1 233,3 tok/s this runAMD 7.0.0-27-generic 147,8 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 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 0.075

Token / kWh4.01M

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)9.84B

☁️ 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-20bNVIDIA GB10 (DGX Spark)	gpt-oss-20bNVIDIA RTX PRO 6000 Blackwell Workstation Edition	gpt-oss-20bNVIDIA RTX PRO 6000 Blackwell Workstation Edition	gpt-oss-20bNVIDIA 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 GB10 (DGX Spark) · this run 156 tok/s NVIDIA RTX PRO 6000 Blackwell Workstation Edition 1,136 tok/s NVIDIA RTX PRO 6000 Blackwell Workstation Edition 1,135 tok/s NVIDIA GeForce RTX 5090 1,114 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.