“More recent models (in particular, Opus 4.8 and Mythos Preview) have “solved” this problem, but only by hardly working together at all: the median agent maintained very high ownership of each of its files, reducing the potential for conflict.”
Quote · anthropic.com
From “Anthropic: newer models avoid PR conflicts by barely sharing files”, found 2026-10-26.
Conflict-free can mean no collaboration. A rising merge rate is not evidence of coordination if code sharing falls to near zero; track both. Early models committed to the same files and abandoned conflicting PRs (Sonnet 4.6 and Opus 4.6 opened 876 and 980 PRs at 80 agents and closed few). Only Sonnet 5 kept both a high merge fraction and real code sharing. Prescriptive-role and CEO-hierarchy prompts made little difference.
- Plan
- Coordinate
- Integrate
AI-drafted