After a Week of Testing, Sonnet 5.5 Wins at Collaboration but Sonnet 5 Still Matches It on Coding
every · x · 2026-09-29
After a week of testing Sonnet 5.5, Every's team published model-selection advice:
- Upgrade if you build iteratively with short feedback rounds, need help outlining arguments or planning projects, or can start at medium effort and check work as it goes
- Stay put if you code: in Kieran Klaassen's 15-task coding test, Sonnet 5 passed 10 tasks at low effort; Sonnet 5.5 passed 9 at low, medium, and high effort—the upgrade didn't improve that score
- Wait if you need unattended, ship-ready builds or polished slide decks—Sonnet 5.5 tends to overbuild and its layouts still needed fixes
- Keep alternatives handy: Opus 5.5 for detailed final builds and longer, harder jobs; GPT-6 Astra for browser-heavy agent work
More from coding & agent
- Dev ships browser-based Fallout: New York built with AI — free to play, just 6MB — chrisfirst · 2026-09-29
- Claude Opus 5.5 runs an AI game studio: builders, critics, revisers, verifiers — chrisfirst · 2026-09-29
- 45 parallel modules, ~580 agent runs: engineering details of an AI game studio — chrisfirst · 2026-09-29
- After a week with Claude Sonnet 5.5: 6 tips, from effort settings to surprise bills — every · 2026-09-29
- Re-enable Ultracode in Claude Code's VS Code extension with one settings. line — AdLow1228 · 2026-09-29
- Dave Morin: No Single Super Agent — Every App Will Become an Agent — nbaschez · 2026-09-29