Deep research that chains several AI models and cites every source
Super Search is a research tool for questions that need more than one search result. You type a question and a chain of AI models works on it in turn: Grok drafts an answer with web search, Gemini fact-checks it with Google Search grounding, and Perplexity's Sonar runs a final verification with its own search. An optional Based AI pass adds what the filtered models leave out. Every claim ends up with sources you can click.
You choose how deep the chain runs, from a single fast pass to six levels, and which models take part. The output is a research document broken into claims, each marked with a confidence level and the models that verified, corrected or flagged it. An activity log shows what each model did, and a token counter shows what the run cost.
You can export the report, copy it as Markdown or text, or share a public link. Upload a CSV, PDF, TXT or Markdown file to have its claims fact-checked, and set a token budget so a run stops when you want it to. It is a premium feature: buying any token package unlocks it.
What you can do
- Chained pipeline: Grok drafts with web search, Gemini fact-checks, Sonar verifies
- Optional Based AI pass through Rhodes that adds what filtered models omit
- Chain depth from one fast pass to six, with any model toggled on or off
- Every claim carries a confidence level, sources, and the models that verified it
- Upload a CSV, PDF, TXT or Markdown file to fact-check its claims
- Export the report, copy Markdown, share a public link, or set a token budget
How it works
- Type your research question and pick the chain depth and the models you want to run.
- Watch the activity log as the drafter writes and each verifier proposes edits.
- Read the research document, grouped by confidence, with sources on every claim.
- Export the report, copy it, or share a public link to it.
Frequently asked questions
Which models does Super Search use?
Grok drafts the answer with web search, Gemini fact-checks it using Google Search grounding, and Perplexity's Sonar performs the final verification with its built-in search. Rhodes, the Based AI engine, can be added as a pass without web search that adds what the filtered models omitted.
How is this different from a single AI chat?
Each claim is checked by more than one model, and the document records which model verified, corrected or flagged it. Claims are grouped by confidence, and sources are attached, so you can see what was checked and how.
Can I fact-check my own document?
Yes. Upload a CSV, PDF, TXT or Markdown file and the chain checks its claims, producing a claim report with status, confidence, the original text, any correction and the sources found.
What does a run cost?
Runs are billed in tokens and the cost depends on the chain depth and the models selected. The page shows tokens used as it goes, and you can set a budget so the run returns a partial result when it is reached.
Do I need to buy tokens to use it?
Super Search is a premium feature. Purchasing any token package unlocks it along with the other premium features. Free tokens cover basic features such as chat and image generation.