Platform matrix
ChatGPT / OpenAI
Allow OAI-SearchBot. Keep public content available without authentication or JavaScript-only rendering. GPTBot is a separate training control.
Gemini / Google
Prioritize Googlebot and Google Search indexing. Google-Extended is a separate control for Gemini model development and grounding, not a Search ranking signal.
Claude
Allow Claude-SearchBot for search discovery and Claude-User for user-directed retrieval. ClaudeBot is the separate model-development crawler.
Perplexity
Allow PerplexityBot. Publish authorship, corrections and first-party expertise — Perplexity says its source review looks at domain-level trust signals.
Grok / xAI
Grok searches the broader web and X with citations. xAI does not currently publish a site-owner robots token comparable to OAI-SearchBot, so broad crawlability and strong web/X entity signals matter.
DeepSeek
DeepSeek offers server-side web search but does not currently publish a dedicated site-owner crawler token. Keep pages accessible to standard search/retrieval systems.
Alice / Yandex
Get the page indexed by Yandex and do not block YandexAdditional/YandexAdditionalBot. Yandex says structured, informative pages are more likely to support Alice answers.
GigaChat
GigaChat offers internet search, but no dedicated webmaster crawler token is currently documented. Strong Russian-language indexing, clear facts and public source pages are the practical foundation.
On-page checklist
- One canonical URL per page and clean EN/RU hreflang.
- Descriptive title, H1 and first paragraph that answer the page’s actual question.
- Visible author and update date on expert material.
- Primary source links where facts depend on external platforms.
- Stable Organization and Person entity IDs in Schema.org.
- Service pages that explain scope, deliverables, limitations and evidence.
- Server logs monitored for crawler access and unexpected 403/429/5xx responses.
Do not optimize for a hallucinated crawler
If a vendor has not documented a crawler token, adding a guessed User-agent name does not create visibility. The fallback is standards-compliant crawlability plus strong indexing in the search ecosystems the agent can access.