llm-benchmarks
githubdrose.io

primeintellect Provider Benchmarks


Measured over the last 30 days

Models Tracked
14
Avg Tokens / Second
40.49
Avg Time to First Token (ms)
6114.29
Last Updated
Nov 18, 2026

All primeintellect models

14 live · one row per model · fastest first · 7 folded behind their current lane
ModelAvg Toks/SecMinMaxAvg TTF (ms)
Moondream 3 Preview86.506.48120.00810.00
Moondream 3.182.5023.90120.00700.00
gpt-oss-20b80.0053.20118.00680.00
DeepSeek V4 Flash 073160.4043.9074.601000.00
gpt-oss-120b52.5049.3055.70970.00
Gemma 4 26B A4B48.4025.1081.70690.00
Ripple 65B Preview33.1030.2036.00800.00
Nemotron 3.5 Super VL31.509.7375.502890.00
Gemma 4 31B30.105.6152.70590.00
Qwen3.6 35B A3B22.804.3758.708560.00
Qwen3.5-9B17.3017.3017.302840.00
DeepSeek V4.1 Flash 101715.405.5337.205370.00
deepseek-v4.1-flash3.973.973.9715800.00
Qwen3.8 27B2.430.778.2143900.00

Frequently Asked Questions

Which primeintellect model is fastest?
Based on recent tests, Moondream 3 Preview shows the highest average throughput among tracked primeintellect models.
How many recent measurements feed this dashboard?
This provider summary aggregates 2149 individual prompts measured across 1850 monitoring runs over the past month.