AEO in July 2026: Opus 5, Voice Mode Citations, and the SE Ranking Study

What changed this week

1. Claude Opus 5 launched (July 24). Adaptive thinking default changes how Claude web search selects citations — reasons about source authority before extracting. Early observation: high-authority technical sources getting cited more, low-authority listicles less.

2. Claude voice mode expanded (July 30). Voice queries surface a different citation pattern than text queries. Shorter, more conversational, requires answer-first content even more strongly. Speakable schema now worth implementing.

3. The SE Ranking llms.txt study. 300K-domain study found llms.txt has “essentially no direct impact” on AI citation rates. Still ship it — for agent routing — but don’t count on it for citation lift.

Sources

• Anthropic — Claude Opus 5 announcement
• Anthropic voice mode expansion
• INITE AI analysis of SE Ranking’s 300K-domain llms.txt study

What to do

→ Verify FAQPage and Article schema on your highest-value pages → Add speakable schema to your top FAQ content → Ship llms.txt (routing benefit, not citation benefit) → Track citation rates in Claude specifically for Opus 5 impact