Forge 1.0
An agentic coding benchmark developed by LeanZero, based on tasks that involve building Atlassian Forge applications in offline environments. It is used to evaluate the agentic programming capabilities of AI models.
Timeline (last 90 days)
On the Forge 1.0 benchmark, Haiku scored 0.9661 and Luna scored 0.395, with Haiku trailing Claude Opus 5.5 by only 0.0106.
Luna never imported historical changes during the Forge test—none of the 62 historical changes were included—triggering a backfill check labeled as critical, which multiplied the entire round's score by 0.6, changing it from 0.6583 to 0.395.
All Facts (2)
On the Forge 1.0 benchmark, Haiku scored 0.9661 and Luna scored 0.395, with Haiku trailing Claude Opus 5.5 by only 0.0106.
50%UnverifiedLuna never imported historical changes during the Forge test—none of the 62 historical changes were included—triggering a backfill check labeled as critical, which multiplied the entire round's score by 0.6, changing it from 0.6583 to 0.395.
50%