Perplexity vs Notebooklm

Official support lists Standard/Free, individual Pro, individual Max, Enterprise Pro, and Enterprise Max. Max includes everything in Pro plus the highest access to advanced models, Comet Max Assistant, Create files and apps, and Brain. The reviewed public consumer-plan pages did not expose current Pro or Max dollar prices. An official organization-onboarding article lists Enterprise Pro at $40/seat monthly or $400/seat annually.

Perplexity is documented as an AI-powered answer/search product with citations, premium models, file uploads, Research, Create files and apps, Comet, and Computer features depending on plan. AI Data Retention is enabled by default for Free, Pro, and Max users and can be turned off for future collection; previously collected training data is not removed by a later opt-out. Enterprise data is not used for AI training, and Perplexity says its third-party model-provider agreements prohibit provider training on Perplexity data. Consumer terms warn that output may be subject to third-party model licenses and use restrictions. 2026-09-02 official blog “Optimizing On-Device Inference for Apple Silicon” (https://www.perplexity.ai/hub/blog/optimizing-on-device-inference-for-apple-silicon): Lily is a lightweight local inference engine built specifically for Apple silicon and Qwen3.6-35B-A3B so on-device compute does not bottleneck Hybrid Compute / Perplexity Computer tasks. Official facts from that post: Rust runtime; OpenAI-compatible chat-completions API; custom Metal kernels; neither PyTorch nor MLX in the execution path; standalone demo on GitHub. The official pplx-garden README names the demo at https://github.com/perplexityai/pplx-garden (lily/), Apache-2.0, greedy generation for the Qwen3.6-35B-A3B MLX affine 4-bit checkpoint format, Apple GPU family 10 or later (M5 and newer). Lily is not a new paid Perplexity plan SKU. Consumer Pro/Max dollar prices were not re-verified on 2026-09-02. Re-verified 2026-10-01: the official plan-comparison article still routes consumer Pro and Max to the app rather than publishing dollar amounts, and lists Education Pro at $10/mo with SheerID verification. Enterprise Pro is stated as starting at $40/month or $400/year per seat, matching the 2026-08-25 record.

NotebookLM is now Gemini Notebook (the official FAQ states the rename took effect in July 2026). The product remains free for individuals with standard usage limits and up to 50 sources per notebook. Gemini Notebook is also included in Google AI Plus (double the standard usage limits, up to 100 sources), Google AI Pro (four times the standard limits, up to 300 sources), and Google AI Ultra (5-20x the Pro limits, up to 600 sources). The plans page does not publish US-dollar prices; the paid tiers are billed through Google One and are available only in listed regions. Organizations, schools, and enterprise access route through Google Workspace or Google Cloud.

Official about page checked 2026-09-01. The page states: 'We value your privacy and never use your organization's data to train Gemini Notebook.' When used as an individual, data is not used for training unless the user shares feedback. The product is powered by Gemini's multimodal capabilities. A support page (support.google.com/gemininotebook/answer/16179536) provides additional privacy details. Privacy and Terms are governed by Google's policies at policies.google.com. Re-verified 2026-10-01: Google now publishes a dedicated Gemini Notebook plans page, which the 2026-09-01 check on the about page did not surface. The plans page is entitlement-based and still publishes no dollar amounts; pricing sits with the Google One AI Plus/Pro/Ultra subscriptions. The privacy statement is unchanged: organization data is never used to train Gemini Notebook, and individual data is not used for training unless the user shares feedback.