LLMON: An LLM-Native Markup Language, now in Granite 4.1 3B and 8B
Which summary reads better? Pick one — models revealed after.Both summaries are AI-generated.
Granite 4.1 now has experimental 3B and 8B preview variants trained to consume and generate LLMON, an XML-like markup that uses reserved special-token delimiters instead of JSON-style punctuation. For production agent builders, the practical shift is that prompt structure, instruction binding, tool/data payload boundaries, and anti-injection scoping can be made explicit to small models rather than inferred from brittle JSON or chat-template conventions.
Granite 4.1 models now support LLMON, a markup language that explicitly communicates structure and semantics to LLMs using reserved special-token delimiters, reducing brittleness and improving robustness against prompt injection attacks; this enables more reliable processing of complex context with multiple components, such as instructions and data payloads, in production LLM applications.