llm config

[fzb408]

If the LLM configuration is not set, then the entire process won't involve the LLM, right?

[hafezparast]

Yes, exactly right! By default, crawl4ai doesn't involve any LLM. The only places where an LLM gets called are:

1. LLMExtractionStrategy — only if you explicitly set extraction_strategy=LLMExtractionStrategy(llm_config=...) in your CrawlerRunConfig
2. LLMContentFilter — only if you pass it as content_filter to your markdown generator
3. AdaptiveCrawler — only when using embedding_llm_config / query_llm_config

Without any of these, you still get raw HTML, cleaned HTML, markdown output, extracted links, and metadata — all processed locally with no API calls.

If you need structured extraction without LLM, check out JsonCssExtractionStrategy or CssExtractionStrategy — they use CSS selectors and are completely LLM-free.