Research notes
Code typing speed benchmarks
A fair way to compare scores by language and duration while representative data is being prepared.
Published 25 September 2026 · Codestroke
“Good typing speed” depends on what you typed and how the test was measured. Codestroke currently publishes eligible Top 20 leaderboard scores, which are useful for competition but are not a representative average for developers.
What Codestroke publishes today
The leaderboard displays top eligible scores by language and test duration. A high rank tells you where a strong validated run sits among other top runs. It does not tell you the median score for everyone who practiced.
Compare like with like
First match the language. Python, SQL, and C++ snippets place different demands on punctuation and line structure. Then match the duration: a 30-second run and a 120-second run answer different questions about pace. Finally, read accuracy beside CPM.
| Keep fixed | Why it matters |
|---|---|
| Language | Syntax and punctuation patterns differ. |
| Duration | Short sprints and longer tests measure different pacing. |
| Scoring rules | Whitespace handling and correction rules can change the result. |
For a personal baseline, run several tests with the same setup and look at the middle result alongside accuracy. This reduces the influence of one unusually easy snippet or one distracted run.
What a public benchmark needs
A useful future dataset should show the period covered, the number of eligible participants, the language and duration for each group, and a distribution such as the median and middle range. It should explain how repeat sessions, suspicious runs, and different input modes were handled. Small groups should be withheld rather than presented as precise population estimates.
When those checks can be met with enough validated data, this page can add actual distributions. Until then, use your own repeated sessions as the most reliable reference for improvement and the leaderboard as a view of top scores.
See how CPM and WPM are calculated before comparing a score from a different tool.
Set your own baseline