# [HISTORICAL] rubric10_semantic scores — publication exemplar run 2026-07-22 (claude-fable-5)
# These score the archived 2026-04-10_sonnet-4.6 label, set aside
# as unattestable. Do not cite as current. See #286.

[claudecode_agent]
  AI_READI   37/50  (74.0%)
  CHORUS     36/50  (72.0%)
  CM4AI      41/50  (82.0%)
  VOICE      44/50  (88.0%)

[claudecode_agent_core]
  AI_READI   44/50  (88.0%)
  CHORUS     42/50  (84.0%)
  CM4AI      43/50  (86.0%)
  VOICE      47/50  (94.0%)

[gpt5]
  AI_READI   21/50  (42.0%)

Note: claudecode_agent and claudecode_agent_core rows evaluated 2026-07-22
with claude-fable-5; other methods retain earlier evaluations (see each JSON's model field).
