Haiku 5.5 max leads on quality and cost: 18.4 points higher and 1.8× cheaper; Opus 4.8 leads on agent time: 2.5× faster.
Stet score
Haiku 5.5 maxRank 6 of 14
58.695% range 47.5–69.018.4 points higher
Opus 4.8Rank 11 of 14
40.295% range 30.3–50.3
Haiku 5.5 max solved 3 tasks that Opus 4.8 didn’t, and Opus 4.8 solved 3 that Haiku 5.5 max didn’t. Both solved 9; neither solved 8.
stet-commit-f8ae1177Passed testsstet-2450ca2dPassed testsstet-commit-342d0ac0Equivalent to the merged fixstet-232b50baPassed testsstet-66762b4cPassed testsstet-43493c52Passed testsA task counts as solved when its tests passed or a judge found the fix equivalent to the merged change, as in the leaderboard’s Solved column. Stet score also weighs code quality, so it can order the two differently. 1 of the 24 tasks is left out because one of the two has no scored result on it.
| Task | Haiku 5.5 max | Opus 4.8 | Solved by |
|---|---|---|---|
stet-3169aefa | Passed tests | Passed tests | Both |
stet-89dfbc27 | Passed tests | Passed tests | Both |
stet-b28f4fe7 | Passed tests | Passed tests | Both |
stet-232b50ba | Failed | Passed tests | Opus 4.8 only |
stet-5276aa3f | Passed tests | Passed tests | Both |
stet-commit-bbbbae09 | Passed tests | Passed tests | Both |
stet-15439c21 | Passed tests | Passed tests | Both |
stet-66762b4c | Failed | Passed tests | Opus 4.8 only |
stet-commit-f0caded9 | Passed tests | Equivalent to the merged fix | Both |
stet-commit-5d766dde | Equivalent to the merged fix | Equivalent to the merged fix | Both |
stet-43493c52 | Failed | Passed tests | Opus 4.8 only |
stet-6f84e978 | Equivalent to the merged fix | Passed tests | Both |
stet-commit-f8ae1177 | Passed tests | Failed | Haiku 5.5 max only |
stet-2450ca2d | Passed tests | Failed | Haiku 5.5 max only |
stet-commit-342d0ac0 | Equivalent to the merged fix | Failed | Haiku 5.5 max only |
stet-commit-5e075641 | Failed | Failed | Neither |
stet-commit-2826765e | Failed | Failed | Neither |
stet-commit-a8716b88 | Failed | Failed | Neither |
stet-commit-30506f8e | Failed | Failed | Neither |
stet-commit-87814583 | Failed | Failed | Neither |
stet-commit-d2ab0c25 | Failed | Failed | Neither |
stet-commit-e41c2e2d | Failed | Failed | Neither |
stet-commit-e928166f | Failed | Failed | Neither |
| Metric | Haiku 5.5 max | Opus 4.8 |
|---|---|---|
| Rank | 6 of 14 | 11 of 14 |
| Stet score | 58.695% range 47.5–69.0 | 40.295% range 30.3–50.3 |
| Solved | 12/249 passed, 3 judged equivalent | 13/2411 passed, 2 judged equivalent |
| Passed review | 9/222 not recorded | 8/231 not recorded |
| Matched merged change | 10/222 not recorded | 11/24 |
| Cost / task | $3.85 | $7.05 |
| How cost was recorded | reported by its harness on 23, estimated from tokens on 1 | reported by its harness |
| Agent time / task | 58.4 min | 23.3 min |
| Ran in | Daytona sandbox | local Docker |
The leaderboard’s readings on groups of tasks. Each group holds fewer tasks than the overall standings, so treat these as leads to test on your repo.
11 tasks
11 tasks whose merged change was 2–149 lines in 1–2 files. The 7 small fixes below are among them.
| Haiku 5.5 max | Opus 4.8 | |
|---|---|---|
| Matched merged change | 7/101 not recorded | 9/11 |
| Passed review | 7/101 not recorded | 8/11 |
| Cost / task | $1.18 | $1.41 |
| Agent time / task | 36.3 min | 12.8 min |
7 tasks
7 tasks whose instructions start with “Fix”. Their merged changes were 4–29 lines in 1–2 files.
Opus 4.8 matched the merged change more often: 7 of 7 to Haiku 5.5 max’s 5.
| Haiku 5.5 max | Opus 4.8 | |
|---|---|---|
| Matched merged change | 5/7 | 7/7 |
| Passed review | 5/7 | 7/7 |
| Cost / task | $1.15 | $1.17 |
| Agent time / task | 34.6 min | 9.7 min |
13 tasks
13 tasks whose merged change ran over 150 lines or more than 3 files: 113–1,420 lines in 2–19 files.
| Haiku 5.5 max | Opus 4.8 | |
|---|---|---|
| Matched merged change | 3/121 not recorded | 2/13 |
| Passed review | 2/121 not recorded | 0/121 not recorded |
| Cost / task | $6.11 | $11.83 |
| Agent time / task | 77.1 min | 32.2 min |
Stet replays your merged PRs, scores quality above pass/fail, and delivers comparison reports through your agent.
Run it on your repocompare-n24-r3b.