llm-benchmarks
githubdrose.io

azure Provider Benchmarks


Measured over the last 30 days

Models Tracked
19
Avg Tokens / Second
26.23
Avg Time to First Token (ms)
5311.05
Last Updated
Oct 16, 2026

All azure models

19 live · one row per model · fastest first
azure site
ModelAvg Toks/SecMinMaxAvg TTF (ms)
GPT-6 Luna Pro90.0055.90115.0020100.00
Claude Haiku 4.543.0043.0043.00610.00
DeepSeek V4 Pro 042338.6010.5052.101600.00
DeepSeek V4 Flash 042338.209.3359.102810.00
GPT-6 Sol Pro33.8011.6049.1012100.00
Claude Sonnet 530.3028.3033.701090.00
Claude Sonnet 4.621.3020.2022.401460.00
Claude Haiku 5.520.6014.9029.102780.00
claude-opus-4.820.6019.8021.401450.00
Claude Opus 4.720.3019.7020.901350.00
GPT-6.1 Sol Pro19.503.7329.9023600.00
Claude Sonnet 4.519.2018.1020.401800.00
Claude Opus 4.618.9017.1020.701890.00
Claude Opus 4.518.5018.3018.801930.00
Claude Haiku Latest15.9015.9015.903430.00
GPT-6 Luna14.103.6122.305080.00
GPT-6 Sol13.505.4921.105030.00
Claude Sonnet 5.511.801.7717.805450.00
GPT-6.1 Sol10.300.6016.107350.00

Frequently Asked Questions

Which azure model is fastest?
Based on recent tests, GPT-6 Luna Pro shows the highest average throughput among tracked azure models.
How many recent measurements feed this dashboard?
This provider summary aggregates 892 individual prompts measured across 843 monitoring runs over the past month.