llm-benchmarks
githubdrose.io

bedrock Provider Benchmarks


Measured over the last 30 days

Models Tracked
22
Avg Tokens / Second
60.15
Avg Time to First Token (ms)
770.45
Last Updated
Aug 9, 2026

All bedrock models

22 tracked · fastest first
ModelAvg Toks/SecMinMaxAvg TTF (ms)
nova-micro120.0027.20188.00310.00
llama-4-maverick108.0023.10135.00290.00
llama-3.1-8b96.007.07112.00400.00
llama-4-scout95.6043.50130.00290.00
nova-lite91.001.05129.00370.00
nova-pro90.8010.80132.00390.00
llama-3.3-70b85.204.60122.00370.00
mistral-7b82.0025.1087.00190.00
llama-3-8b77.801.0586.70260.00
mixtral-8x7b73.1019.3080.60240.00
mistral-small56.2023.0060.60210.00
llama-3.1-70b52.802.0073.60480.00
llama-3.2-90b46.8036.2049.90350.00
claude-haiku-4.543.301.0459.00930.00
mistral-large42.105.7546.60260.00
llama-3-70b36.301.0441.50550.00
claude-opus-4-727.801.0338.701710.00
claude-sonnet-4.627.006.3633.10980.00
claude-sonnet-4.522.001.0029.301510.00
claude-opus-4.621.003.5824.801690.00
claude-opus-4.520.101.0523.301660.00
Claude Opus 4.18.411.1415.703510.00

Frequently Asked Questions

Which bedrock model is fastest?
Based on recent tests, nova-micro shows the highest average throughput among tracked bedrock models.
How many recent measurements feed this dashboard?
This provider summary aggregates 29005 individual prompts measured across 26464 monitoring runs over the past month.