Skip to the comparison
STET
Get a free trial

GPT 6.1 Sol medium vs Haiku 5.5 max

GPT 6.1 Sol medium leads on quality, cost and agent time: 6.7 points higher, 3.5× cheaper and 5.0× faster.

24 tasksData from

Stet score

GPT 6.1 Sol mediumRank 2 of 14

65.395% range 55.6–74.66.7 points higher

Haiku 5.5 maxRank 6 of 14

58.695% range 47.5–69.0

A directional read: their 95% score ranges overlap.

Cost / task

GPT 6.1 Sol medium$1.103.5× cheaper
Haiku 5.5 max$3.85

Agent time / task

GPT 6.1 Sol medium11.8 min5.0× faster
Haiku 5.5 max58.4 min

Task by task

GPT 6.1 Sol medium solved 3 tasks that Haiku 5.5 max didn’t, and Haiku 5.5 max solved 2 that GPT 6.1 Sol medium didn’t. Both solved 10; neither solved 8.

GPT 6.1 Sol medium only 3

  • stet-232b50baPassed tests
  • stet-66762b4cPassed tests
  • stet-43493c52Passed tests

Haiku 5.5 max only 2

  • stet-commit-f0caded9Passed tests
  • stet-2450ca2dPassed tests

A 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.

All 23 compared tasks
Task by task
TaskGPT 6.1 Sol mediumHaiku 5.5 maxSolved by
stet-3169aefaPassed testsPassed testsBoth
stet-89dfbc27Passed testsPassed testsBoth
stet-b28f4fe7Passed testsPassed testsBoth
stet-232b50baPassed testsFailedGPT 6.1 Sol medium only
stet-5276aa3fPassed testsPassed testsBoth
stet-commit-bbbbae09Passed testsPassed testsBoth
stet-15439c21Passed testsPassed testsBoth
stet-66762b4cPassed testsFailedGPT 6.1 Sol medium only
stet-commit-f0caded9FailedPassed testsHaiku 5.5 max only
stet-commit-5d766ddeEquivalent to the merged fixEquivalent to the merged fixBoth
stet-43493c52Passed testsFailedGPT 6.1 Sol medium only
stet-6f84e978Equivalent to the merged fixEquivalent to the merged fixBoth
stet-commit-f8ae1177Passed testsPassed testsBoth
stet-2450ca2dFailedPassed testsHaiku 5.5 max only
stet-commit-342d0ac0Equivalent to the merged fixEquivalent to the merged fixBoth
stet-commit-5e075641FailedFailedNeither
stet-commit-2826765eFailedFailedNeither
stet-commit-a8716b88FailedFailedNeither
stet-commit-30506f8eFailedFailedNeither
stet-commit-87814583FailedFailedNeither
stet-commit-d2ab0c25FailedFailedNeither
stet-commit-e41c2e2dFailedFailedNeither
stet-commit-e928166fFailedFailedNeither

Side by side

Side by side
MetricGPT 6.1 Sol mediumHaiku 5.5 max
Rank2 of 146 of 14
Stet score65.395% range 55.6–74.658.695% range 47.5–69.0
Solved13/2410 passed, 3 judged equivalent12/249 passed, 3 judged equivalent
Passed review10/249/222 not recorded
Matched merged change10/2410/222 not recorded
Cost / task$1.10$3.85
How cost was recordedestimated from tokensreported by its harness on 23, estimated from tokens on 1
Agent time / task11.8 min58.4 min
Ran inDaytona sandboxDaytona sandbox

Where each setup is the better choice

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.

Changes up to 150 lines

11 tasks

11 tasks whose merged change was 2–149 lines in 1–2 files. The 7 small fixes below are among them.

Changes up to 150 lines
GPT 6.1 Sol mediumHaiku 5.5 max
Matched merged change7/117/101 not recorded
Passed review8/117/101 not recorded
Cost / task$0.47$1.18
Agent time / task5.8 min36.3 min

Small fixes

7 tasks

7 tasks whose instructions start with “Fix”. Their merged changes were 4–29 lines in 1–2 files.

GPT 6.1 Sol medium matched the merged change more often: 6 of 7 to Haiku 5.5 max’s 5.

Small fixes
GPT 6.1 Sol mediumHaiku 5.5 max
Matched merged change6/75/7
Passed review5/75/7
Cost / task$0.42$1.15
Agent time / task5.3 min34.6 min

Larger changes

13 tasks

13 tasks whose merged change ran over 150 lines or more than 3 files: 113–1,420 lines in 2–19 files.

Larger changes
GPT 6.1 Sol mediumHaiku 5.5 max
Matched merged change3/133/121 not recorded
Passed review2/132/121 not recorded
Cost / task$1.63$6.11
Agent time / task16.9 min77.1 min

From the leaderboard’s routing advice

  • Small fix: to spend less, use GPT 6.1 Sol medium.
  • Larger change: use GPT 6.1 Sol medium.
  • Anything else: use GPT 6.1 Sol medium.

See which setup wins on your repo

Stet replays your merged PRs, scores quality above pass/fail, and delivers comparison reports through your agent.

Run it on your repo

About this data

  • 24 tasks, each a merged change replayed from the repo as it was just before the merge.
  • Data from .
  • Both setups come from the same leaderboard run of 14 setups: compare-n24-r3b.

See all 14 setups on the leaderboardHow Stet measures