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

gemma-4-E2B-it

Performance benchmark · measured on 27.07.2026 17:54

Benchmark-IDrun-20260727-181606-0adf6d
Timebench 3 - Kombi (Prefill + Generation)Dense5BRuntime: llama.cppQuantisierung: Q4_K_M
Generation1.349,34tok/s
Prefill18.159,46tok/s
Time to First Token1.301,50ms
Total duration14,89s
Concurrency5parallel
Ranking in the field
1of 20 systems

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

This run is better than 100 % of all comparable systems.
Generation 1.349,3 tok/s
+139 % vs Ø 564,9
Prefill 18.159,5 tok/s
+17 % vs Ø 15.471,6
Time to First Token 1.302 ms
-91 % vs Ø 14.022
Distribution in the field0 – 1.349 tok/s
Ø 565 Median Dieser Lauf

Wie schlägt sich dieser Benchmark mit anderen Modellen?

gemma-4-E2B-it this runNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260727-181606-0adf6d
1.349,3 tok/s
NVIDIA-Nemotron-3-Nano-4BNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260727-181607-30d864
1.188,7 tok/s
gemma-4-E4B-itNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260727-181607-bf4219
914,3 tok/s
gpt-oss-20bNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260723-194447-1679c5
842,9 tok/s
Qwen3-Coder-30B-A3B-InstructNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260724-004605-81e02e
676,8 tok/s
Qwen3-VL-30B-A3B-InstructNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260724-004607-0c918b
666,5 tok/s
Qwen3-30B-A3B-Instruct-2507NVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260724-004604-c65af3
649,0 tok/s
Qwen3-30B-A3B-Thinking-2507NVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260724-004605-09fb68
625,7 tok/s
Qwen3-Omni-30B-A3B-ThinkingNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260724-004611-280a45
618,2 tok/s
Ministral-3-14B-Reasoning-2512NVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260723-195749-bde180
615,8 tok/s
Qwen3.6-35B-A3BNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260724-083000-a4b89b
596,5 tok/s
Qwen3.5-35B-A3BNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260724-082959-b75c95
594,8 tok/s
Devstral-Small-2-24B-Instruct-2512NVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260724-004603-7a14f9
423,4 tok/s
Magistral-Small-2509NVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260724-004604-d66137
411,8 tok/s

How does this benchmark compare on other GPUs?

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

Hardware

GPU: NVIDIA RTX PRO 6000 Blackwell Workstation Edition · 96 GB VRAM
CPU: AMD Ryzen 9 9950X 16-Core Processor
RAM: 92 GB
Mainboard: ASUSTeK COMPUTER INC. ProArt X870E-CREATOR WIFI

Setup

Runtime: llama.cpp
Quantization: Q4_K_M
Model: gemma-4-E2B-it

Configuration

benchmark-konfiguration — run-20260727-181606-0adf6d
# LLM-Benchmark Konfiguration # Modell : gemma-4-E2B-it # Engine : llama.cpp # Run-ID : run-20260727-181606-0adf6d # 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--unsloth--gemma-4-E2B-it-GGUF/snapshots/0314792d7f1f7e229411f620751375812bb9faf2/gemma-4-E2B-it-Q4_K_M.gguf \ --alias gemma-4-E2B-it \ --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.gemma-4-E2B-it
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--unsloth--gemma-4-E2B-it-GGUF/snapshots/0314792d7f1f7e229411f620751375812bb9faf2/gemma-4-E2B-it-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

Model comparison

gemma-4-E2B-it 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

3.0852.3141.5437710,03157.27514.23421.193Prefill (tok/s)Generation (tok/s)NVIDIA GeForce RTX 5070 Ti - 1.743,1 tok/s Generation, 17.651 tok/s Prefill, TTFT 2.033 ms (5 Laufe)NVIDIA GeForce RTX 50...NVIDIA GeForce RTX 3090 Ti - 1.492,8 tok/s Generation, 9.978 tok/s Prefill, TTFT 3.182 ms (3 Laufe)NVIDIA GeForce RTX 30...NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition - 266,1 tok/s Generation, 3.857 tok/s Prefill, TTFT 15.559 ms (3 Laufe)NVIDIA RTX PRO 6000 B...NVIDIA RTX PRO 6000 Blackwell Workstation Edition - 2.414,4 tok/s Generation, 16.315 tok/s Prefill, TTFT 1.871 ms (3 Laufe) | DIESER LAUF★ NVIDIA RTX PRO 6000 B...
★ NVIDIA RTX PRO 6000 Blackwell Workstation Edition 2.414,4 tok/s this runNVIDIA GeForce RTX 5070 Ti 1.743,1 tok/sNVIDIA GeForce RTX 3090 Ti 1.492,8 tok/sNVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition 266,1 tok/s

CPUby processor

3.0852.3141.5437710,03157.27514.23421.193Prefill (tok/s)Generation (tok/s)AMD Ryzen Threadripper PRO 5975WX 32-Cores - 1.743,1 tok/s Generation, 17.651 tok/s Prefill, TTFT 2.033 ms (5 Laufe)AMD Ryzen Threadrippe...AMD Ryzen 9 8945HX with Radeon Graphics - 1.492,8 tok/s Generation, 9.978 tok/s Prefill, TTFT 3.182 ms (3 Laufe)AMD Ryzen 9 8945HX wi...AMD Ryzen Threadripper PRO 9965WX 24-Cores - 266,1 tok/s Generation, 3.857 tok/s Prefill, TTFT 15.559 ms (3 Laufe)AMD Ryzen Threadrippe...AMD Ryzen 9 9950X 16-Core Processor - 2.414,4 tok/s Generation, 16.315 tok/s Prefill, TTFT 1.871 ms (3 Laufe) | DIESER LAUF★ AMD Ryzen 9 9950X 16-...
★ AMD Ryzen 9 9950X 16-Core Processor 2.414,4 tok/s this runAMD Ryzen Threadripper PRO 5975WX 32-Cores 1.743,1 tok/sAMD Ryzen 9 8945HX with Radeon Graphics 1.492,8 tok/sAMD Ryzen Threadripper PRO 9965WX 24-Cores 266,1 tok/s

MBby mainboard

3.0852.3141.5437710,03157.27514.23421.193Prefill (tok/s)Generation (tok/s)ASUSTeK COMPUTER INC. Pro WS WRX80E-SAGE SE WIFI - 1.743,1 tok/s Generation, 17.651 tok/s Prefill, TTFT 2.033 ms (5 Laufe)ASUSTeK COMPUTER INC....Meigao Innovation Technology (Shen Zhen) Co., Ltd DRFXL (MotherBoard Series) - 1.492,8 tok/s Generation, 9.978 tok/s Prefill, TTFT 3.182 ms (3 Laufe)Meigao Innovation Tec...ASUSTeK COMPUTER INC. Pro WS WRX90E-SAGE SE - 266,1 tok/s Generation, 3.857 tok/s Prefill, TTFT 15.559 ms (3 Laufe)ASUSTeK COMPUTER INC....ASUSTeK COMPUTER INC. ProArt X870E-CREATOR WIFI - 2.414,4 tok/s Generation, 16.315 tok/s Prefill, TTFT 1.871 ms (3 Laufe) | DIESER LAUF★ ASUSTeK COMPUTER INC....
★ ASUSTeK COMPUTER INC. ProArt X870E-CREATOR WIFI 2.414,4 tok/s this runASUSTeK COMPUTER INC. Pro WS WRX80E-SAGE SE WIFI 1.743,1 tok/sMeigao Innovation Technology (Shen Zhen) Co., Ltd DRFXL (MotherBoard Series) 1.492,8 tok/sASUSTeK COMPUTER INC. Pro WS WRX90E-SAGE SE 266,1 tok/s

ENGby engine

2.8172.4142.0111.6091.2066.78713.02119.25525.489Prefill (tok/s)Generation (tok/s)vLLM - 1.608,5 tok/s Generation, 22.042 tok/s Prefill, TTFT 757 ms (3 Laufe)vLLMllama.cpp - 2.414,4 tok/s Generation, 10.235 tok/s Prefill, TTFT 6.339 ms (11 Laufe) | DIESER LAUF★ llama.cpp
★ llama.cpp 2.414,4 tok/s this runvLLM 1.608,5 tok/s

DRVby driver

2.6562.5352.4142.2942.17311.99912.51013.02013.531Prefill (tok/s)Generation (tok/s)unbekannt - 2.414,4 tok/s Generation, 12.765 tok/s Prefill, TTFT 5.143 ms (14 Laufe)unbekannt
unbekannt 2.414,4 tok/s
💰 Economics

Economics of this run

Operating cost, TCO and comparison with the next-best runs of the same model at identical concurrency (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 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 0.040
Token / kWh7.55M
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)85.11B
☁️ 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

gemma-4-E2B-itNVIDIA RTX PRO 6000 Blackwell Workstation Editiongemma-4-E2B-itNVIDIA GeForce RTX 5070 Tigemma-4-E2B-itNVIDIA GeForce RTX 5070 Tigemma-4-E2B-itNVIDIA GeForce RTX 3090 Ti
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.