LLM Runtime Status

This page is retained for existing links. LLM-facing work is not currently an active public capability claim.

Active Repository Surface

• Optional ml feature dependencies exist in aether-lang.
• Interpreter value variants and native function names include LLM-related entries.
• Tensor, neural, and ML helper primitives exist in aether-core.

Gated Surface

Do not claim LLM inference performance, model quality, memory reduction, or hardware acceleration without:

• a runnable command;
• model name and artifact hash;
• hardware and software environment;
• prompt/input set;
• correctness or output-quality check;
• baseline implementation;
• timing and memory artifact.

See Benchmark Policy.