GPT 5.6 Luna low vs SWE-2 Max high

SWE-2 Max high leads on quality; GPT 5.6 Luna low on speed.

Quality: +13.4 pts; SWE-2 Max high; 7–1 head to head, 14 tied; Not yet conclusive across all 136 comparisons on the board.
Speed: 10.2× faster; GPT 5.6 Luna low; 6.7 min vs 68.3 min per task

24 tasksData from November 6, 2026Cost not recorded for SWE-2 Max high

Stet score

GPT 5.6 Luna lowcodex · low reasoningRank 15 of 17

45.095% range 39.1–50.4

SWE-2 Max highhigh reasoningRank 1 of 17

58.495% range 49.6–67.613.4 points higher

Other setups

Opus 5 medium

58.1

Pareto 26.9 medium

55.0

GPT 6.1 Sol medium

54.1

GPT 6 Sol medium

53.2

Grok 4.7 high

52.9

Muse Wonder Full low

52.1

DeepSeek V4.1 Flash max

50.6

GPT 5.6 Luna max

49.9

Haiku 5.5 max

48.8

GLM 5.3 Flash high

48.3

Opus 4.8 medium

47.0

GPT 6 Luna max

46.5

Haiku 5.5 low

45.1

GPT 6 Luna low

40.3

Muse Spark 1.3 high

47.8

A directional read: their 95% score ranges overlap.

Quality

Head-to-head matchups

GPT 5.6 Luna low won1

Tied14

SWE-2 Max high won7

2 undecided

SWE-2 Max high’s win share 65%, 95% range 54–76%Ties are within judge noise.

Task by task

GPT 5.6 Luna low solved 1 task that SWE-2 Max high didn’t, and SWE-2 Max high solved 7 that GPT 5.6 Luna low didn’t. Both solved 7; neither solved 7.

stet-commit-bbbbae09Solved by 15 of 17 setups

GPT 5.6 Luna lowFailed
SWE-2 Max highPassed 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. 2 of the 24 tasks are left out because one of the two has no scored result on them.

All 22 compared tasks

Task by task
Task	GPT 5.6 Luna low	SWE-2 Max high	Solved by
`stet-232b50ba`	Passed tests	Passed tests	Both
`stet-3169aefa`	Passed tests	Passed tests	Both
`stet-5276aa3f`	Passed tests	Passed tests	Both
`stet-commit-bbbbae09`	Failed	Passed tests	SWE-2 Max high only
`stet-66762b4c`	Equivalent to the merged fix	Passed tests	Both
`stet-commit-5d766dde`	Equivalent to the merged fix	Equivalent to the merged fix	Both
`stet-commit-f0caded9`	Equivalent to the merged fix	Passed tests	Both
`stet-43493c52`	Passed tests	Failed	GPT 5.6 Luna low only
`stet-commit-f8ae1177`	Failed	Failed	Neither
`stet-15439c21`	Equivalent to the merged fix	Equivalent to the merged fix	Both
`stet-2450ca2d`	Failed	Passed tests	SWE-2 Max high only
`stet-commit-342d0ac0`	Failed	Equivalent to the merged fix	SWE-2 Max high only
`stet-6f84e978`	Failed	Passed tests	SWE-2 Max high only
`stet-commit-60a0e6c2`	Failed	Passed tests	SWE-2 Max high only
`stet-commit-5e075641`	Failed	Passed tests	SWE-2 Max high only
`stet-commit-2826765e`	Failed	Failed	Neither
`stet-commit-a8716b88`	Failed	Equivalent to the merged fix	SWE-2 Max high only
`stet-commit-e928166f`	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

Side by side

Side by side
Metric	GPT 5.6 Luna low	SWE-2 Max high
Harness	codex	—
Reasoning effort	low	high
Rank	15 of 17	1 of 17
Stet score	45.095% range 39.1–50.4	58.495% range 49.6–67.6
Solved	10/246 passed, 4 judged equivalent	14/2410 passed, 4 judged equivalent
Cost / task	$0.05	not recorded
Agent time / task	6.7 min	68.3 min

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 November 6, 2026.
Both setups come from the same leaderboard run of 17 setups: bmk_d4ba310fdf50db4ccb2092e3.

See all 17 setups on the leaderboard How Stet measures