Measured performance
AMD Ryzen AI 9 365 / Linux. Timings are medians in milliseconds; lower is better.
Ready to edit
These measurements compare editing readiness with isolated configurations and LSP disabled. They do not compare equivalent feature setups: Runyte includes integrated terminals, Git workflows, and persistent sessions, while other editors provide different combinations of built-in features and plugins. Results do not predict performance with your configuration.
Time from launch to a visible edit, with warm caches and LSP disabled. Initial syntax parsing may still be running. Ten measured launches per cell follow one warm-up; every sample passes whole-file save verification.
30 September 2026 · Runyte v0.3.5 (b937570) · Neovim 0.12.5 · Helix 25.07.1
| File | Lines | Neovim | Helix | Runyte |
|---|---|---|---|---|
| Text | 500 | 20.9 | 32.2 | 18.8 |
| Text | 5,000 | 20.9 | 32.7 | 20.4 |
| Text | 50,000 | 22.6 | 33.0 | 26.9 |
| Text | 500,000 | 38.6 | 45.9 | 88.0 |
| Lua | 500 | 42.3 | 39.1 | 20.6 |
| Lua | 5,000 | 31.7 | 63.7 | 21.7 |
| Lua | 50,000 | 31.3 | 349.0 | 27.9 |
| Lua | 500,000 | 45.7 | 548.8 | 102.5 |
Detailed results and methodology on GitHub
Persistent sessions
Runyte time to visible output, median of three samples per scenario.
30 September 2026 · Runyte v0.3.5 (b937570)
| Scenario | Cold start | Warm attach |
|---|---|---|
| One session | 34.23 | 6.12 |
| Three sessions | 33.26 | 6.80 |
| Three sessions, strip hidden | 34.04 | 6.27 |
| Three sessions, another terminal producing output | 33.43 | 6.70 |
Session results and methodology on GitHub
Quit and idle
Quit medians use ten launches. Idle uses five ten-second windows with a 5,000-line Lua file open in a Git repository.
30 September 2026 · Runyte v0.3.5 (b937570)
| Measurement | Runyte |
|---|---|
| Quit immediately after first document frame, 50,000-line Lua | 4.9 ms |
| Quit after settling, 50,000-line text | 4 ms |
| Quit after settling, 50,000-line Lua | 22 ms |
| Idle CPU, median | 0.00% |
| Idle screen writes | 0 |