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

Mistral-Small-3.1-24B-Instruct-2503

Performance benchmark · measured on 21.08.2026 15:16

Benchmark-IDrun-20260821-152103-31cc7c
Timebench 3 - Kombi (Prefill + Generation)Dense24BRuntime: llama.cppQuantisierung: Q4_K_M
Generation23,84tok/s
Prefill277,02tok/s
Time to First Token48.246,50ms
Total duration528,72s
Concurrency5parallel
Ranking in the field
679of 841 systems

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

This run is better than 19 % of all comparable systems.
Generation 23,8 tok/s
-91 % vs Ø 274,3
Prefill 277,0 tok/s
-93 % vs Ø 3.891,1
Time to First Token 48.247 ms
+15 % vs Ø 41.842
Distribution in the field0 – 1.349 tok/s
Ø 274 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 GeForce RTX 5090 · run-20260728-184455-5b4937
1.301,9 tok/s
gemma-4-E2B-it3× 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
gpt-oss-20bNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260728-140954-9e81f9
1.136,3 tok/s
gpt-oss-20bNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260730-035052-09627f
1.135,0 tok/s
gpt-oss-20bNVIDIA GeForce RTX 5090 · run-20260729-032121-058a31
1.113,5 tok/s
Nemotron-3-Nano-4B3× NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260728-194136-836a91
1.090,6 tok/s
Nemotron-3-Nano-4B3× NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260727-181606-8b381e
1.077,4 tok/s
Nemotron-3-Nano-4B3× NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260728-194136-b8a818
1.071,0 tok/s
Nemotron-3-Nano-4B3× NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260728-194136-038e92
1.051,1 tok/s
gpt-oss-20b3× NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260729-032121-ffb18a
1.015,5 tok/s
Nemotron-3-Nano-30B-A3BNVIDIA RTX PRO 6000 Blackwell Workstation Edition · run-20260729-032121-305ad6
1.007,8 tok/s
Mistral-Small-3.1-24B-Instruct-2503 this runNVIDIA Tesla P100 PCIe 16GB · run-20260821-152103-31cc7c
23,8 tok/s

How does this benchmark compare on other GPUs?

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

Configuration

benchmark-konfiguration — run-20260821-152103-31cc7c
# LLM-Benchmark Konfiguration # Modell : Mistral-Small-3.1-24B-Instruct-2503 # Engine : llama.cpp # Run-ID : run-20260821-152103-31cc7c # GPU : NVIDIA Tesla P100 PCIe 16GB # CPU : AMD Ryzen 9 7945HX with Radeon Graphics # RAM : 60 GB bench@llm-benchmark:~$ /home/godcore/llama.cpp/build/bin/llama-server \ -m mistral-small-3.1-24b-instruct-2503-hf-imat-Q4_K_M.gguf \ -ngl 999 \ -fa on \ -c 24000 \ -np 5
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
Kontextlaenge?Maximale Anzahl Tokens (Eingabe + erzeugte Ausgabe zusammen), die das Modell pro Anfrage verarbeiten kann.24000
Modellpfad?Pfad zur GGUF-Modelldatei, die geladen und ausgeliefert wird.mistral-small-3.1-24b-instruct-2503-hf-imat-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
faon
Kontext?Groesse des Kontextfensters in Token. 0 = der beim Training verwendete Kontext des Modells.24000
np5

All benchmarks of this model To leaderboard

Anzeige
Model comparison

Mistral-Small-3.1-24B-Instruct-2503 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

19114395,447,70,003.4356.87110.306Prefill (tok/s)Generation (tok/s)AMD Radeon PRO W7900 Dual Slot - 150,4 tok/s Generation, 1.883 tok/s Prefill, TTFT 8.173 ms (12 Laufe)AMD Radeon PRO W7900 ...NVIDIA RTX PRO 6000 Blackwell Workstation Edition - 88,3 tok/s Generation, 8.343 tok/s Prefill, TTFT 1.652 ms (3 Laufe)NVIDIA RTX PRO 6000 B...AMD Radeon AI PRO R9700 - 33,1 tok/s Generation, 3.057 tok/s Prefill, TTFT 1.231 ms (16 Laufe)AMD Radeon AI PRO R97...AMD Radeon 8060S Graphics - 31,3 tok/s Generation, 1.069 tok/s Prefill, TTFT 21.414 ms (2 Laufe)AMD Radeon 8060S Grap...NVIDIA Tesla P100 PCIe 16GB - 23,8 tok/s Generation, 218 tok/s Prefill, TTFT 30.869 ms (2 Laufe) | DIESER LAUF★ NVIDIA Tesla P100 PCI...
AMD Radeon PRO W7900 Dual Slot 150,4 tok/sNVIDIA RTX PRO 6000 Blackwell Workstation Edition 88,3 tok/sAMD Radeon AI PRO R9700 33,1 tok/sAMD Radeon 8060S Graphics 31,3 tok/s★ NVIDIA Tesla P100 PCIe 16GB 23,8 tok/s this run

CPUby processor

19114395,447,70,003.4356.87110.306Prefill (tok/s)Generation (tok/s)AMD Ryzen Threadripper PRO 5975WX 32-Cores - 150,4 tok/s Generation, 1.883 tok/s Prefill, TTFT 8.173 ms (12 Laufe)AMD Ryzen Threadrippe...AMD Ryzen 9 9950X 16-Core Processor - 88,3 tok/s Generation, 8.343 tok/s Prefill, TTFT 1.652 ms (3 Laufe)AMD Ryzen 9 9950X 16-...AMD Ryzen Threadripper PRO 7955WX 16-Cores - 33,1 tok/s Generation, 3.057 tok/s Prefill, TTFT 1.231 ms (16 Laufe)AMD Ryzen Threadrippe...AMD RYZEN AI MAX+ 395 w/ Radeon 8060S - 31,3 tok/s Generation, 1.069 tok/s Prefill, TTFT 21.414 ms (2 Laufe)AMD RYZEN AI MAX+ 395...AMD Ryzen 9 7945HX with Radeon Graphics - 23,8 tok/s Generation, 218 tok/s Prefill, TTFT 30.869 ms (2 Laufe) | DIESER LAUF★ AMD Ryzen 9 7945HX wi...
AMD Ryzen Threadripper PRO 5975WX 32-Cores 150,4 tok/sAMD Ryzen 9 9950X 16-Core Processor 88,3 tok/sAMD Ryzen Threadripper PRO 7955WX 16-Cores 33,1 tok/sAMD RYZEN AI MAX+ 395 w/ Radeon 8060S 31,3 tok/s★ AMD Ryzen 9 7945HX with Radeon Graphics 23,8 tok/s this run

MBby mainboard

19114395,447,70,003.4356.87110.306Prefill (tok/s)Generation (tok/s)ASUSTeK COMPUTER INC. Pro WS WRX80E-SAGE SE WIFI - 150,4 tok/s Generation, 1.883 tok/s Prefill, TTFT 8.173 ms (12 Laufe)ASUSTeK COMPUTER INC....ASUSTeK COMPUTER INC. ProArt X870E-CREATOR WIFI - 88,3 tok/s Generation, 8.343 tok/s Prefill, TTFT 1.652 ms (3 Laufe)ASUSTeK COMPUTER INC....ASUSTeK COMPUTER INC. Pro WS WRX90E-SAGE SE - 33,1 tok/s Generation, 3.057 tok/s Prefill, TTFT 1.231 ms (16 Laufe)ASUSTeK COMPUTER INC....Bosgame AXB35-02 (BeyondMax Series) - 31,3 tok/s Generation, 1.069 tok/s Prefill, TTFT 21.414 ms (2 Laufe)Bosgame AXB35-02 (Bey...Shenzhen Meigao Electronic Equipment Co.,Ltd F1FXM (DeskMini Series) - 23,8 tok/s Generation, 218 tok/s Prefill, TTFT 30.869 ms (2 Laufe) | DIESER LAUF★ Shenzhen Meigao Elect...
ASUSTeK COMPUTER INC. Pro WS WRX80E-SAGE SE WIFI 150,4 tok/sASUSTeK COMPUTER INC. ProArt X870E-CREATOR WIFI 88,3 tok/sASUSTeK COMPUTER INC. Pro WS WRX90E-SAGE SE 33,1 tok/sBosgame AXB35-02 (BeyondMax Series) 31,3 tok/s★ Shenzhen Meigao Electronic Equipment Co.,Ltd F1FXM (DeskMini Series) 23,8 tok/s this run

ENGby engine

18914294,447,20,02392.8975.5548.211Prefill (tok/s)Generation (tok/s)vLLM - 88,3 tok/s Generation, 6.853 tok/s Prefill, TTFT 1.087 ms (6 Laufe)vLLMunbekannt - 33,1 tok/s Generation, 3.057 tok/s Prefill, TTFT 1.231 ms (8 Laufe)unbekanntllama.cpp - 150,4 tok/s Generation, 1.597 tok/s Prefill, TTFT 10.044 ms (21 Laufe) | DIESER LAUF★ llama.cpp
★ llama.cpp 150,4 tok/s this runvLLM 88,3 tok/sunbekannt 33,1 tok/s

DRVby driver

18914294,447,20,02.5292.7843.0383.293Prefill (tok/s)Generation (tok/s)unbekannt - 150,4 tok/s Generation, 2.765 tok/s Prefill, TTFT 8.053 ms (27 Laufe)unbekanntAMD 7.0.0-27-generic - 33,1 tok/s Generation, 3.057 tok/s Prefill, TTFT 1.231 ms (8 Laufe)AMD 7.0.0-27-generic
unbekannt 150,4 tok/sAMD 7.0.0-27-generic 33,1 tok/s
💰 Economics

Economics of this run

Operating cost, TCO and comparison with the next-best runs of the same model at identical concurrency (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 0 W
⚡ TDP 250 W
☁️ External LLM (API)
Electricity0.30 EUR/kWh
Avg power (incl. idle)250 W estimated (TDP)GPU 250 W full load
Avg cost / hourEUR 0.075
Electricity / 1M tokensEUR 0.87
Token / kWh343.30K
Acquisition (system)EUR 120 missingPSU EUR 120
Electricity (2 years)
TCO (2 years)EUR 1,434
Output tokens (2 years)1.50B
☁️ 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 (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

Mistral-Small-3.1-24B-Instruct-2503NVIDIA Tesla P100 PCIe 16GBMistral-Small-3.1-24B-Instruct-2503AMD Radeon PRO W7900 Dual SlotMistral-Small-3.1-24B-Instruct-25032x AMD Radeon PRO W7900 Dual SlotMistral-Small-3.1-24B-Instruct-25032x AMD Radeon PRO W7900 Dual Slot
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.