Zapier's AI Default Is a 5× Billing Trap

Zapier rebranded its Agents product as “AI by Zapier” on October 6, 2026, folding all agent capabilities into a standard AI step inside the Zap editor. The architecture is cleaner. The billing is a trap. Every new AI step on a paid plan defaults to the Premium model tier — 5× your normal task rate — and every tool call the AI makes gets multiplied at the same rate. Build a Zap with one AI step and three tool calls at the Premium default and you’re burning 20 tasks per run, not one. That’s not a footnote. That’s a budget-destroying default.

What Happened

On June 15, 2026, Zapier introduced model-tier pricing for AI steps inside the Zap editor. Three tiers: Standard (1× tasks), Advanced (3×), Premium (5×). The June 15 launch already defaulted new steps to Advanced at 3× — not Standard. That was already expensive. On August 20, 2026, Zapier pushed the default further, to Premium at 5×. That change got minimal fanfare. Most users didn’t notice until their task counts spiked.

The October 6 rebrand makes the damage universal. Before October 6, model-tier pricing only hit users who had opted into the AI step inside the editor. Now that Agents is folded into AI by Zapier as the sole AI path, every Zapier user building AI automations is subject to the 5× default — not just early adopters.

The multiplier math is what makes this genuinely punishing. Zapier’s own documentation confirms: a Premium AI step that makes three tool calls costs (1 × 5) + (3 × 5) = 20 tasks per run. A Professional plan user at $19.99/month gets 750 tasks. Run that 20-task Zap 38 times and you’ve hit the ceiling. Run it daily and you’re out of tasks in 37 days, not a full month.

Zapier frames the consolidation as a simplification — one history, one editor, one place to control cost and capability per step. That part is true. The problem is the default. Zapier set it to the most expensive option and left it to users to find the setting and change it manually on every step they build.

Why It Matters

This isn’t a price increase to a new tier you can opt out of. It’s a default that burns through your existing task allocation faster than anything you’ve built before. And the users who get hit hardest are exactly the users most excited to use AI automation: freelancers and small-team operators on the Professional plan, running experiments and iterating fast.

A Professional plan user ($19.99/month, 750 tasks) building an AI-assisted lead qualification Zap — one AI step, two tool calls — burns 15 tasks per run at Premium default. At 100 runs a month, that’s 1,500 tasks against a 750-task ceiling. They hit the wall halfway through the month without realizing the AI step is the culprit.

A Team plan user ($69/month, 2,000 tasks) has more headroom but still runs into trouble at scale. One documented case from a user processing 100 leads daily with an 8-step AI workflow reached $847/month in Zapier costs before they caught the multiplier and switched.

The second-order effect is subtler: because the default is Premium, users who build AI Zaps and don’t check the model tier setting are training themselves to see AI automation as expensive. The cost isn’t inherent to AI steps — it’s inherent to the default Zapier picked.

The Take

Change the default on every AI step you’ve already built. Open the AI by Zapier step settings, find the model tier selector, and switch it to Standard (1×) unless you have a specific reason to need the more capable model. Standard doesn’t support tool calls — if your AI step needs to call other apps, you’re looking at Advanced (3×) as the minimum. If you bring your own API key, Zapier bills the step at 1× regardless of which model you’re calling; you pay your AI provider directly for model usage.

Those mitigations work. But they require you to know the trap exists, find the setting, and change it on every new step you build. That’s ongoing vigilance, not a one-time fix.

If AI automation is a meaningful part of your workflow — not a curiosity, but a regular part of what you bill time against — Make Core at $9/month annual is worth pricing out right now. Make bills AI module calls in credits. No per-run multipliers on basic model calls. 10,000 credits/month on the Core plan. For a freelancer running 500 AI-assisted automation runs a month, Make’s credit model doesn’t compound the way Zapier’s task multiplier does. The cost difference can exceed $30–50/month at that volume, and it only widens as you scale.

Zapier still wins on integration depth — 9,000+ apps versus Make’s narrower native library — and the new unified editor is genuinely better than the old Agents interface. If you’re running low-volume AI automations (under 100 runs/month) on a Team plan, the multiplier doesn’t bite hard enough to force a migration. Stay, but switch your defaults.

If you’re on Professional and actively building AI Zaps, the math is not in Zapier’s favor.