A measure of how accurately a language model adheres to explicit constraints, formatting schemas, and rules provided in a prompt.

Marketed by model providers as a breakthrough obedience metric deserving special accolades. It takes baseline expectations for functional software—that it follow explicit formatting instructions without making stuff up—and presents it as a revolutionary behavioral milestone.

• Obedience Framing: Treating standard code execution and string formatting like a well-behaved house pet.
• The Compliance Mirage: Claiming "superior instruction-following" right up until a subtle prompt tweak causes the software to ignore output boundaries.
• Baseline Expectation as Feature: Selling basic JSON output compliance as an advanced AI capability.

Ferris Bueller's Day Off (1980s) scene where Grace checks Ed Rooney's computerized attendance log, carefully explaining how Ferris meticulously follows rules on paper while actively orchestrating a wild day off across Chicago.

How accurately a software tool follows specific formatting rules and instructions provided in a prompt.

“Selecting models with advanced instruction-following capabilities guarantees strict schema compliance.”

“The software follows custom formatting instructions and output rules more accurately than older versions.”