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

Ornith-1.5-9B

Performance benchmark · measured on 12.11.2026 15:59

Benchmark-IDrun-20261112-152252-1e6a9b
Timebench 3 - Kombi (Prefill + Generation)Dense9BRuntime: lmstudio-mlxQuantisierung: MLX 4bit (OptiQ)
Generation21,94tok/s
Prefill1.046,83tok/s
Time to First Token2.009,20ms
Total duration149,54s
Concurrency1parallel
Ranking in the field
1449of 1927 systems

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

This run is better than 25 % of all comparable systems.
Generation 21,9 tok/s
-72 % vs Ø 77,5
Prefill 1.046,8 tok/s
-63 % vs Ø 2.844,6
Time to First Token 2.009 ms
-92 % vs Ø 23.755
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, ARM64, AdvSIMD)
RAM: 24 GB
Mainboard: Apple Mac mini

Setup

Runtime: lmstudio-mlx
Quantization: MLX 4bit (OptiQ)
Model: Ornith-1.5-9B

Configuration

benchmark-konfiguration — run-20261112-152252-1e6a9b
# LLM-Benchmark Konfiguration # Modell : Ornith-1.5-9B # Engine : lmstudio-mlx # Run-ID : run-20261112-152252-1e6a9b # GPU : Keine GPU (CPU-only) # CPU : Apple M6 (unified CPU/GPU SoC, ARM64, AdvSIMD) # RAM : 24 GB bench@llm-benchmark:~$ lmstudio-mlx \ --model ornith-1.5-9b-optiq
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.lmstudio-mlx
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

Ornith-1.5-9B 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

6094573051520,01911.8633.5345.206Prefill (tok/s)Generation (tok/s)NVIDIA RTX A6000 - 472,2 tok/s Generation, 4.350 tok/s Prefill, TTFT 3.813 ms (12 Laufe)NVIDIA RTX A6000AMD Radeon AI PRO R9700 - 197,0 tok/s Generation, 3.694 tok/s Prefill, TTFT 2.761 ms (3 Laufe)AMD Radeon AI PRO R97...CPU-only - 21,9 tok/s Generation, 1.047 tok/s Prefill, TTFT 2.009 ms (1 Lauf) | DIESER LAUF★ CPU-only
NVIDIA RTX A6000 472,2 tok/sAMD Radeon AI PRO R9700 197,0 tok/s★ CPU-only 21,9 tok/s this run

CPUby processor

6094573051520,01911.8633.5345.206Prefill (tok/s)Generation (tok/s)AMD Ryzen Threadripper PRO 7955WX 16-Cores - 472,2 tok/s Generation, 4.350 tok/s Prefill, TTFT 3.813 ms (12 Laufe)AMD Ryzen Threadrippe...AMD Ryzen 3 3100 4-Core Processor - 197,0 tok/s Generation, 3.694 tok/s Prefill, TTFT 2.761 ms (3 Laufe)AMD Ryzen 3 3100 4-Co...Apple M6 (unified CPU/GPU SoC, ARM64, AdvSIMD) - 21,9 tok/s Generation, 1.047 tok/s Prefill, TTFT 2.009 ms (1 Lauf) | DIESER LAUF★ Apple M6 (unified CPU...
AMD Ryzen Threadripper PRO 7955WX 16-Cores 472,2 tok/sAMD Ryzen 3 3100 4-Core Processor 197,0 tok/s★ Apple M6 (unified CPU/GPU SoC, ARM64, AdvSIMD) 21,9 tok/s this run

MBby mainboard

6094573051520,01911.8633.5345.206Prefill (tok/s)Generation (tok/s)ASUSTeK COMPUTER INC. Pro WS WRX90E-SAGE SE - 472,2 tok/s Generation, 4.350 tok/s Prefill, TTFT 3.813 ms (12 Laufe)ASUSTeK COMPUTER INC....ASUSTeK COMPUTER INC. ROG STRIX X570-F GAMING - 197,0 tok/s Generation, 3.694 tok/s Prefill, TTFT 2.761 ms (3 Laufe)ASUSTeK COMPUTER INC....Apple Mac mini - 21,9 tok/s Generation, 1.047 tok/s Prefill, TTFT 2.009 ms (1 Lauf) | DIESER LAUF★ Apple Mac mini
ASUSTeK COMPUTER INC. Pro WS WRX90E-SAGE SE 472,2 tok/sASUSTeK COMPUTER INC. ROG STRIX X570-F GAMING 197,0 tok/s★ Apple Mac mini 21,9 tok/s this run

ENGby engine

6094573051520,02231.8293.4365.043Prefill (tok/s)Generation (tok/s)llama.cpp - 472,2 tok/s Generation, 4.219 tok/s Prefill, TTFT 3.603 ms (15 Laufe)llama.cpplmstudio-mlx - 21,9 tok/s Generation, 1.047 tok/s Prefill, TTFT 2.009 ms (1 Lauf) | DIESER LAUF★ lmstudio-mlx
llama.cpp 472,2 tok/s★ lmstudio-mlx 21,9 tok/s this run

DRVby driver

5194964724494253.7793.9404.1014.262Prefill (tok/s)Generation (tok/s)unbekannt - 472,2 tok/s Generation, 4.021 tok/s Prefill, TTFT 3.503 ms (16 Laufe)unbekannt
unbekannt 472,2 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)1.38B
☁️ 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

Ornith-1.5-9BKeine GPU (CPU-only)Ornith-1.5-9B2x NVIDIA RTX A6000Ornith-1.5-9B2x NVIDIA RTX A6000Ornith-1.5-9B2x NVIDIA RTX A6000
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.