Codex
codexCodex's pass rate on every board it ran on. Scores on different boards don't add up, so this page doesn't add them.
On each board
Terminal-Bench 2.0
terminal-bench@2.0GPT-5.6 Sol0.163.0, full, 89 tasks, 1 attempt, open
62.9%52.8–73.0
humaneval-plus@1.0 (unreviewed)
Every board here: 0.164.0, full, 164 tasks, 1 attempt, openRanks 1–2
2 harnesses95.1%91.5–98.2Codex minus Claude Code 2.1.309: 0.0 points, 95% interval −1.8 to +1.8, over 164 tasks both ran; too close to call.
2 harnesses95.1%91.5–98.2Codex minus Claude Code 2.1.309: 0.0 points, 95% interval −1.8 to +1.8, over 164 tasks both ran; too close to call.
Ranks 1–2
2 harnesses94.5%90.9–97.6Codex 0.164.0 is the reference.
2 harnesses94.5%90.9–97.6Codex 0.164.0 is the reference.
Ranks 1–2
2 harnesses95.7%92.7–98.8Codex 0.164.0 is the reference.
2 harnesses95.7%92.7–98.8Codex 0.164.0 is the reference.
Ranks 1–2
2 harnesses95.7%92.7–98.2Codex 0.164.0 is the reference.
2 harnesses95.7%92.7–98.2Codex 0.164.0 is the reference.
Ranks 1–3
3 harnesses95.7%92.7–98.2Codex minus DeepSeek Harness 0.2.1-rc.2: −0.6 points, 95% interval −1.8 to 0.0, over 164 tasks both ran; too close to call.
3 harnesses95.7%92.7–98.2Codex minus DeepSeek Harness 0.2.1-rc.2: −0.6 points, 95% interval −1.8 to 0.0, over 164 tasks both ran; too close to call.
Ranks 1–2
2 harnesses95.7%92.7–98.2Codex minus Claude Code 2.1.309: −0.6 points, 95% interval −1.8 to 0.0, over 164 tasks both ran; too close to call.
2 harnesses95.7%92.7–98.2Codex minus Claude Code 2.1.309: −0.6 points, 95% interval −1.8 to 0.0, over 164 tasks both ran; too close to call.
Ranks 1–2
2 harnesses88.4%83.5–93.3Codex minus Claude Code 2.1.309: −4.3 points, 95% interval −9.1 to 0.0, over 164 tasks both ran; too close to call.
2 harnesses88.4%83.5–93.3Codex minus Claude Code 2.1.309: −4.3 points, 95% interval −9.1 to 0.0, over 164 tasks both ran; too close to call.