Created bymario-alka.dePowered bynoob2claw.detricoma.de
Contributed byMario AlkaGoogle

gemma-3n-E4B-it

Performance benchmark · measured on 12.11.2026 10:05

Benchmark-IDrun-20261112-092948-cab409
Timebench 3 - Kombi (Prefill + Generation)Dense8BRuntime: OllamaQuantisierung: Q4_K_M
Generation39,36tok/s
Prefill837,78tok/s
Time to First Token2.387,50ms
Total duration92,46s
Concurrency1parallel
Ranking in the field
1100of 1897 systems

Performance benchmark · Primary metric: Generation-Speed (tok/s) · 1× concurrent

This run is better than 42 % of all comparable systems.
Generation 39,4 tok/s
-49 % vs Ø 77,8
Prefill 837,8 tok/s
-71 % vs Ø 2.850,0
Time to First Token 2.388 ms
-90 % vs Ø 24.083
Distribution in the field0 – 471 tok/s
Ø 78 Median Dieser Lauf

Wie schlägt sich dieser Benchmark mit anderen Modellen?

How does this benchmark compare on other GPUs?

Same model on different hardware · 1× concurrent · Generation (tok/s)

Hardware

GPU: Keine GPU (CPU-only)
CPU: Apple M6 (unified CPU/GPU SoC)
RAM: 24 GB
Mainboard: Apple Mac mini

Setup

Runtime: Ollama
Quantization: Q4_K_M
Model: gemma-3n-E4B-it

Configuration

benchmark-konfiguration — run-20261112-092948-cab409
# LLM-Benchmark Konfiguration # Modell : gemma-3n-E4B-it # Engine : Ollama # Run-ID : run-20261112-092948-cab409 # GPU : Keine GPU (CPU-only) # CPU : Apple M6 (unified CPU/GPU SoC) # RAM : 24 GB bench@llm-benchmark:~$ ollama serve \ --model gemma3n:e4b
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.ollama
Kontextlaenge?Maximale Anzahl Tokens (Eingabe + erzeugte Ausgabe zusammen), die das Modell pro Anfrage verarbeiten kann.8192
execution_typelocal

All benchmarks of this model To leaderboard

Anzeige
Model comparison

gemma-3n-E4B-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

34726017486,80,01421.5032.8644.225Prefill (tok/s)Generation (tok/s)AMD Radeon RX 7900 XTX - 273,1 tok/s Generation, 3.529 tok/s Prefill, TTFT 2.950 ms (9 Laufe)AMD Radeon RX 7900 XTXCPU-only - 39,4 tok/s Generation, 838 tok/s Prefill, TTFT 2.388 ms (1 Lauf) | DIESER LAUF★ CPU-only
AMD Radeon RX 7900 XTX 273,1 tok/s★ CPU-only 39,4 tok/s this run

CPUby processor

34726017486,80,0651.5833.1014.619Prefill (tok/s)Generation (tok/s)AMD Ryzen Threadripper PRO 3955WX 16-Cores - 273,1 tok/s Generation, 3.847 tok/s Prefill, TTFT 2.646 ms (6 Laufe)AMD Ryzen Threadrippe...Intel(R) Xeon(R) CPU E5-4657L v2 @ 2.40GHz - 220,8 tok/s Generation, 2.894 tok/s Prefill, TTFT 3.558 ms (3 Laufe)Intel(R) Xeon(R) CPU ...Apple M6 (unified CPU/GPU SoC) - 39,4 tok/s Generation, 838 tok/s Prefill, TTFT 2.388 ms (1 Lauf) | DIESER LAUF★ Apple M6 (unified CPU...
AMD Ryzen Threadripper PRO 3955WX 16-Cores 273,1 tok/sIntel(R) Xeon(R) CPU E5-4657L v2 @ 2.40GHz 220,8 tok/s★ Apple M6 (unified CPU/GPU SoC) 39,4 tok/s this run

MBby mainboard

34726017486,80,0651.5833.1014.619Prefill (tok/s)Generation (tok/s)ASUSTeK COMPUTER INC. Pro WS WRX80E-SAGE SE WIFI - 273,1 tok/s Generation, 3.847 tok/s Prefill, TTFT 2.646 ms (6 Laufe)ASUSTeK COMPUTER INC....Dell Inc. PowerEdge R820 - 220,8 tok/s Generation, 2.894 tok/s Prefill, TTFT 3.558 ms (3 Laufe)Dell Inc. PowerEdge R...Apple Mac mini - 39,4 tok/s Generation, 838 tok/s Prefill, TTFT 2.388 ms (1 Lauf) | DIESER LAUF★ Apple Mac mini
ASUSTeK COMPUTER INC. Pro WS WRX80E-SAGE SE WIFI 273,1 tok/sDell Inc. PowerEdge R820 220,8 tok/s★ Apple Mac mini 39,4 tok/s this run

ENGby engine

34726017486,80,01421.5032.8644.225Prefill (tok/s)Generation (tok/s)llama.cpp - 273,1 tok/s Generation, 3.529 tok/s Prefill, TTFT 2.950 ms (9 Laufe)llama.cppOllama - 39,4 tok/s Generation, 838 tok/s Prefill, TTFT 2.388 ms (1 Lauf) | DIESER LAUF★ Ollama
llama.cpp 273,1 tok/s★ Ollama 39,4 tok/s this run

DRVby driver

3002872732592463.0643.1953.3253.456Prefill (tok/s)Generation (tok/s)unbekannt - 273,1 tok/s Generation, 3.260 tok/s Prefill, TTFT 2.894 ms (10 Laufe)unbekannt
unbekannt 273,1 tok/s
💰 Economics

Economics of this run

Operating cost, TCO and comparison with the next-best runs of the same model at identical concurrency (1× 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 %
🔌 Idle 0 W
⚡ TDP 10 W
☁️ External LLM (API)
Electricity0.30 EUR/kWh
Avg power (incl. idle)0 W missing
Avg cost / hour–
Electricity / 1M tokens–
Token / kWh–
Acquisition (system)– partial price
Electricity (2 years)–
TCO (2 years)–
Output tokens (2 years)2.48B
☁️ External LLM (API) – comparison
External LLM cost (2 years)–
Savings vs. external (2 years)–
No power draw measured – values estimated from GPU TDP + CPU (idle + 15 %) + board.

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 (0 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-3n-E4B-itKeine GPU (CPU-only)gemma-3n-E4B-itAMD Radeon RX 7900 XTXgemma-3n-E4B-itAMD Radeon RX 7900 XTXgemma-3n-E4B-it2x AMD Radeon RX 7900 XTX
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.