The $0 Research Stack: NotebookLM, Elicit, and Consensus in 2026
Students and analysts can build a serious AI research workflow without paying anything. Here's the stack, and when each paid tier earns its price.
By Editorial team · July 19, 2026 · 1 min read
Research tools are the rare AI category where the free tiers genuinely cover a full workflow. Here's how they fit together.
The free stack
Explore with Perplexity (free: limited Pro searches daily). Twenty minutes of conversation orients you in an unfamiliar topic — treat it as a map, not a source.
Extract with Elicit (free tier included). Semantic search across 125M+ papers, pulling sample sizes, methods, and outcomes into structured tables. The systematic-review grind, automated.
Verify with Consensus (free syntheses monthly). Its Consensus Meter shows how much peer-reviewed literature actually agrees with a claim — the validation layer after exploration.
Synthesize with NotebookLM (free, generously). Upload your 20 best papers; it answers only from them, with citations to the exact passage. The Audio Overviews feature — turning your sources into a podcast-style discussion — remains the most surprising free feature in AI.
When to pay
Consensus Pro at $9/month (annual) is the cheapest meaningful upgrade for anyone doing evidence checks weekly. Elicit's paid tiers ($12-49/month) matter once you run real systematic reviews. Perplexity Pro at $20 makes sense for daily researchers — noting Deep Research is now capped around 20/month. Scite ($12/month) is the citation-integrity layer academics add before submission.
Total for a fully paid stack: roughly $50/month. Total for the version that covers most people: $0.