llm-benchmarks
githubdrose.io

mistral Provider Benchmarks


Measured over the last 30 days

Models Tracked
16
Avg Tokens / Second
47.88
Avg Time to First Token (ms)
1733.75
Last Updated
Oct 14, 2026

All mistral models

16 live · one row per model · fastest first · 1 retired lanes not listed · 3 folded behind their current lane
mistral site
ModelAvg Toks/SecMinMaxAvg TTF (ms)
Ministral 3 3B 251263.8052.2075.50810.00
GLM 5.261.7043.7079.60660.00
Mistral Medium 360.9057.2064.60630.00
Voxtral Small 24B 250758.6055.0062.20680.00
Codestral 250857.8052.7062.80710.00
Saba57.3051.5063.10710.00
Mistral Small 457.0055.4058.50660.00
Mixtral 8x22B Instruct54.0052.2055.80710.00
Ministral 3 14B 251251.5048.1055.00590.00
Mistral Medium 3.148.6045.9050.80760.00
Ministral 3 8B 251247.7046.7048.70780.00
Mistral Medium 3.546.5010.8066.30970.00
Mistral Large 240735.2035.2035.20730.00
Mistral Large 3 251233.6033.2034.00830.00
Mistral Large26.8022.0031.701210.00
Mistral Large 45.080.8310.5016300.00

Frequently Asked Questions

Which mistral model is fastest?
Based on recent tests, Ministral 3 3B 2512 shows the highest average throughput among tracked mistral models.
How many recent measurements feed this dashboard?
This provider summary aggregates 440 individual prompts measured across 426 monitoring runs over the past month.