Real Atlassian solutions to real problems — no fluff, no SEO spam.
The entrant that tops our leaderboard is gpt-5.6-sol running inside goose's own harness. The six days that produced nothing were OpenAI's Codex, a different product. Two variables moved and I controlled one. What survives is narrower and more interesting: on the harder scorer it was killed at a two-hour wall and still placed first, which is a thing a leaderboard cannot show you.
