Factory releases Droid 1.0 agent harness, 65.2% on Terminal-Bench 4 with Claude Opus 5.5

Factory has released Droid 1.0, a model-independent coding agent harness that makes programmatic tool calling the default. On Factory's own Terminal-Bench 4 runs, Droid scores 65.2% with Claude Opus 5.5 at $4.73 per passed task.

Key details

• Product: Droid 1.0, a coding agent harness from Factory.
• Design: Model-independent, meaning the harness is not tied to a single underlying model.
• Default behavior: Programmatic tool calling is the default.
• Foundation: MarkTechPost's headline describes Droid 1.0 as built on Script Mode.
• Benchmark: 65.2% on Terminal-Bench 4 with Claude Opus 5.5, from Factory's own runs.
• Cost: $4.73 per passed task in those runs.

The benchmark and cost figures come from Factory's own Terminal-Bench 4 runs, so they are company-reported results.

Droid 1.0 is a harness, not a new model. Pricing for Droid itself, context window size and other benchmark configurations are not part of the figures reported here.

The details come from a MarkTechPost report published on October 31, 2026.

What this means

Droid 1.0 sits in the agent layer around a model, not in the model itself. Factory's reported result pairs the harness with Claude Opus 5.5, so the 65.2% score reflects that specific combination. The cost per passed task of $4.73 gives a spend figure alongside the pass rate, which matters for teams comparing agent setups on cost as well as accuracy. Because the runs are Factory's own, the numbers are best read as the company's claim about its harness.