llm-benchmarks
githubdrose.io

cloudflare Provider Benchmarks


Measured over the last 30 days

Models Tracked
20
Avg Tokens / Second
17.31
Avg Time to First Token (ms)
9893.60
Last Updated
Sep 1, 2026

All cloudflare models

20 tracked · fastest first
ModelAvg Toks/SecMinMaxAvg TTF (ms)
Llama 3.2 3B Instruct82.9070.4096.70520.00
Llama 3.2 1B Instruct51.7029.9062.90680.00
Llama 3.2 1B Instruct51.1022.60135.001050.00
Gemma 4 26B A4B48.2036.5070.50700.00
Gemma 4 26B A4B43.2033.6051.40750.00
DeepSeek V4 Flash 042336.8029.5049.40850.00
Mistral Small 3.1 24B29.1012.2047.401080.00
Llama 3.3 70B Instruct29.0027.0031.00880.00
Granite 4.0 Micro28.4025.3030.40610.00
GLM 5.228.0016.7046.901980.00
Granite 4.0 Micro25.6021.0033.30910.00
Mistral Small 3.1 24B25.5021.9028.40860.00
Qwen2.5 Coder 32B Instruct25.3023.8026.30610.00
Qwen2.5 Coder 32B Instruct24.1016.7027.10810.00
DeepSeek V4 Flash 073122.4015.1026.102400.00
Llama 3.1 8B Instruct16.2014.9016.80840.00
Kimi K2.7 Code9.787.1013.005810.00
Kimi K2.64.863.776.0112600.00
GLM 4.7 Flash3.643.643.6416100.00
GLM 4.7 Flash2.692.013.3523300.00

Frequently Asked Questions

Which cloudflare model is fastest?
Based on recent tests, Llama 3.2 3B Instruct shows the highest average throughput among tracked cloudflare models.
How many recent measurements feed this dashboard?
This provider summary aggregates 1468 individual prompts measured across 1214 monitoring runs over the past month.