llm-benchmarks
githubdrose.io

anthropic Provider Benchmarks


Measured over the last 30 days

Models Tracked
10
Avg Tokens / Second
25.92
Avg Time to First Token (ms)
1770.00
Last Updated
Sep 22, 2026

All anthropic models

10 live · one row per model · fastest first · 1 folded behind their current lane
anthropic site
ModelAvg Toks/SecMinMaxAvg TTF (ms)
claude-opus-4.8-fast55.8045.2071.90890.00
Claude Haiku 4.542.7040.1045.30800.00
claude-opus-4.835.4022.2048.301090.00
Claude Sonnet 523.4019.9028.201490.00
Claude Opus 4.522.3019.7023.701260.00
Claude Opus 4.720.6016.3026.201370.00
Claude Sonnet 4.620.0014.9023.701670.00
Claude Sonnet 4.519.9019.3020.401530.00
Claude Opus 4.618.3017.7019.601690.00
Claude Fable 50.840.840.845910.00

Frequently Asked Questions

Which anthropic model is fastest?
Based on recent tests, claude-opus-4.8-fast shows the highest average throughput among tracked anthropic models.
How many recent measurements feed this dashboard?
This provider summary aggregates 105 individual prompts measured across 83 monitoring runs over the past month.