Coding · sub-raceNarrow lead
Terminal & DevOps race
Working in a real shell: builds, containers, CI, infrastructure as code and server administration.
- Leader
- Sonnet 5.5
- Margin
- 0.9 pts
- 1.3% ahead
- Weeks at #1
- 4
- in a row
- Changes
- 5
- in 2026
Warming up the track…
Oct 19, 2026Sonnet 5.5 leads
2025Apr ’25Jul ’25Oct ’252026Apr ’26Jul ’26Oct ’26
Score history
Top 6 today over the leader's frontier · click to jump to a week
#1 frontierSonnet 5.5Opus 5.5GPT-6 AstraGPT-6.1 SolFable 5.1Gemini 4 Argon
Reigns at #1
Longest: Fable 5, 15 weeks (from Jun 2026).
Sibling races
More coding races
Why this race matters
Terminal skill is what lets an agent actually ship, not just suggest.
How it is scored
Task success rate in sandboxed terminals.
Unit: percentHigher is betterPhoto finish under 0.5 ptsDominant over 3.0 pts
Data provenance
Hard, sandboxed command-line tasks with automated checks.
Scores here are illustrative placeholders that mimic real benchmark patterns. Swap in live data through the score provider. Methodology