Factory Droid 1.0: Model-Independent Coding Agent
Factory launches Droid 1.0, a model-independent coding agent harness using Script Mode, scoring 65.2% on Terminal-Bench 4.
Stock photo for illustration only, not from the actual event
- Factory introduced Droid 1.0 as a model-independent coding agent harness.
- It establishes programmatic tool calling as the default operating mode.
- Achieved a score of 65.2% on Terminal-Bench 4 using Claude Opus 5.5.
- Maintains an operational cost of $4.73 per successfully passed task.
Factory has officially released Factory Droid 1.0, a model-independent coding agent harness designed to make programmatic tool calling the core default mechanism for software development tasks.
According to Factory's internal evaluations on Terminal-Bench 4, Droid 1.0 achieved a passing score of 65.2% when paired with Claude Opus 5.5. The operational cost recorded during these runs was $4.73 per successfully completed task.
Stock photo for illustration only, not from the actual event
The introduction of a model-independent agent harness represents a significant shift in AI software engineering. By decoupling the agent framework from specific large language models, developers gain the flexibility to integrate newer or alternative foundational models without restructuring their entire tool-calling pipeline.
This release highlights the ongoing industry push toward balancing high benchmark performance with transparent, manageable costs in autonomous coding workflows.
Source: MarkTechPost
Found something wrong in this article? Report an issue with this article
Comments
Leave a Comment