Factory Says Droid 1.0 Tops Codex on Terminal-Bench 4 at 12% Lower Cost
Summary
Factory released Droid 1.0, an agent harness that now runs tool calls through sandboxed JavaScript programs, cutting model requests on long tasks. Factory says Droid scores 57.6% on Terminal-Bench 4 with GPT-6.1 Sol, against 54.0% for Codex, at 12% lower cost. The company claims up to 20% more efficient use than models' native harnesses.
Key Points
- Factory says its Script tool runs JavaScript in a sandbox with no file, network, or system access.
- Factory says Script mode is on by default for ten models, including GPT-6.1 Sol, Claude Opus 5.5, and Kimi K3.
- Factory reports one 312-call Linear cleanup took 2 model requests and roughly 61,000 input tokens.