Quality mode

Quality mode runs the full multi-model pipeline on premium models — research, drafting, critique, claim verification, merge — so the output you ship is the best one-shot piece our system can produce. Use it whenever the content is going to be read by someone outside your team.

Speed vs Quality

Every build picks one of two modes. The toggle lives at the top of the build form and defaults to Speed.

When to choose Quality

• Anything public-facing — blog posts, landing pages, customer emails, white papers, sales decks.
• Anything you can't easily revise — published content, outbound campaigns, executive communications.
• Complex topics — technical deep-dives, regulated industries, or anything where a wrong claim is expensive.
• Pillar / cornerstone content — pieces designed to rank, get cited by AI search, and stay live for months.

When Speed is the right call

• Drafts and exploration — early ideation, brainstorming outlines, rough takes you'll iterate on.
• Internal-only — meeting notes, internal updates, ops documents.
• High-volume runs — when you need ten variants quickly to pick one to invest in.
• Tight time budgets — when you need a publishable draft now and will edit by hand.

What it costs

• Speed: 1 token per page
• Quality: 4 tokens per page
• Quality + Research: 8 tokens per page

The 4× factor reflects the per-call API price difference of premium models — the same delta you'd pay if you ran the pipeline yourself.

How long it takes

• Speed: 30–60 seconds per page
• Quality: 2–5 minutes per page
• Quality + Research: 3–5 minutes per page

The extra time is the price of multiple LLMs running in sequence — research, draft, critique, merge — and chewing through more reasoning tokens at each step.

What you actually get with Quality

Sharper reasoning

• Multi-step arguments hold together end to end
• Counterpoints get acknowledged instead of glossed over
• Structure follows the argument, not a template
• Subtle inconsistencies get caught before they ship

Deeper review

• The critique pass surfaces weak claims that single-model runs miss
• Improvement suggestions are specific and actionable
• Fact-checking accuracy is meaningfully higher

Stronger research

• More sources reviewed per pass
• Source quality is assessed, not just collected
• Synthesis across multiple sources, not paraphrasing one

Pair it with Research mode

For your most important pieces, turn on both Research and Quality. The combination — premium models + real-time grounded sources — is what we recommend for any flagship content.

• Premium models on every step
• Real-time web research with cited sources
• 8 tokens per page
• 3–5 minutes per page

Tips for getting the most out of Quality

• Invest in your objective. Premium models reward specific prompts. Vague objectives waste tokens.
• Don't rush. Premium models think longer. Walk away.
• Review the claim ledger. Quality mode generates more claims and verifies them more strictly — the ledger is where you confirm.
• Compare side-by-side. Run the same objective in both modes once to see the delta on your kind of content.
• Enable email notifications. Quality builds take long enough that you'll want to start them and come back.