AI writing: one model, one lower price
We upgraded the AI writing engine to a single model that outperformed every option on our writing picker in blind testing — and costs less. The writing model picker is gone; everyone now pays the lowest rate on the ladder.
- AI research + writing now runs on GPT-5.6 Luna for everyone. In our blind panel testing (a 10-judge panel over 200 identical leads) it was the strongest researcher of the engines we evaluated — and it costs less to run than the models it replaces, so keeping the more expensive picker options would have meant charging more for worse emails.
- The volume-band ladder was re-derived for the single model, and every band went down: $0.20, $0.16, $0.13, $0.10 and $0.08 per output (was $0.30, $0.25, $0.20, $0.15 and $0.12 at the old mid model — and even campaigns on the old cheapest setting pay the same or less at every band). Bands still work like tax brackets, counted across all your workspaces over a rolling 30 days.
- The included research-and-writing allowance and the overage rates are unchanged, and spintax follow-up variants still cost $0.
- The scoring model picker is unchanged — scoring still offers three models at $1.50 / $4.50 / $7.50 per 1,000 leads.
- This is a price decrease, effective immediately and prospectively: it applies to runs confirmed from today. Per our price-change policy, nothing already confirmed is repriced.