Sonnet 5.5 lands in Factory: early tests show 'High' is the strong default
matanSF · x · 2026-09-29
Agent-native software development company Factory announced that Sonnet 5.5 is now live on its platform, sharing initial observations:
- High reasoning effort is a strong default setting
- The model checks whether the real requirement is met, not just the nearest failing test — a key quality improvement
- Explanations carried over from earlier work, showing better cross-session context retention
Factory's Droid platform lets teams plan, build, review, and ship software with autonomous agents across desktop, terminal, web, and CLI, with built-in model routing, code review, and automation features.
More from coding & agent
- Claude rescues years-old PhD defense slides after JS dependencies broke — Ben_Reinhardt · 2026-09-29
- Indie dev builds a high-performance AI video editor on his own infrastructure — wavefnx · 2026-09-29
- JetBrains launches Air Teams to let whole teams run and improve agentic workflows — shensi · 2026-09-29
- Roque Nights, a Stargazing Planning Agent, Wins WebMCP Challenge — thisiskp_ · 2026-09-29
- Frontier VLMs Still Fail at Parsing Forms; LlamaIndex Ships a Purpose-Built Fix — llama_index · 2026-09-29
- Qwen 3.8 27B as a workhorse subagent with DeepSeek as orchestrator — lordekeen · 2026-09-29