LaunchDevelopersOctober 31, 2026

Factory ships Droid 1.0 coding agent harness with programmatic tool calling

Read original source →marktechpost.com

Base prompt cut 37% to 13,055 input tokens, with only 2 tool definitions up front instead of 11. The model writes sandboxed JavaScript that runs tools, so one request can drive hundreds of calls. Factory reports 65.2% on Terminal-Bench 4 with Claude Opus 5.5 at $4.73 per passed task.

1 source

More stories today

Open the live feed