Factory Droid 1.0 beats Codex and Claude Code in Factory's own tests
Why it matters
Factory released Droid 1.0 (29 October 2026), the biggest update to its model-independent coding agent harness. In Factory's own Terminal-Bench 4 runs (three trials per task, October 23–28), Droid with GPT-6.1 Sol scored 57.6% versus 54.0% for Codex at 12% lower cost, and with Claude Opus 5.5 scored 65.2% versus 62.1% for Cursor and 57.6% for Claude Code. The biggest change is programmatic tool calling: the model writes a JavaScript program that runs hundreds of tool calls in a single request; one 312-call Linear cleanup took 2 requests versus an estimated 36 with direct calls. Available as a desktop app, terminal UI and SDK. Caveat: self-reported benchmarks, not independent results.
Read the source: factory.com (opens in a new tab)
All hits from 30 Oct 2026More CodingGet the daily hits by email