From the source
Factory is releasing Droid 1.0, an update to its agent harness that it says is more accurate, efficient and faster than before, aimed at enterprise coding work.
- The company says Droid outperforms the native harnesses built for their models by up to 20% more efficiently on the same tasks.
- On Terminal-Bench 4 with GPT-6.1 Sol, Factory says Droid scores 3.5 points higher than Codex at 12% lower cost, and 6.6 to 12.6 points higher than OpenCode and Devin.
- With Claude Opus 5.5, it says Droid scores 3.0 to 7.6 points higher than
Cursor, OpenCode, Devin and Claude Code, at up to 24% lower cost.
- The largest change is programmatic tool calling, now the default: the agent writes a JavaScript program that invokes tools in a sandbox with no file, network or system access of its own.
- Factory says this keeps context clean, cuts model requests and makes checks run deterministically in code, and that a Script can make hundreds of tool calls in one model request.
