Free
Token calculator

No-code AI agent builders compared on price (2026)

An AI agent handling 1,000 requests a month fits in Make Core ($12/mo yearly) or n8n Starter (€20/mo yearly). On Zapier it needs a $89-$189/mo plan.

By · Updated

Quick comparison

ProviderFromStarting planFree tierOur rating
MakeChecked
$12/moCore10,000 credits/month; unlimited active scenarios; scheduling down to the minute1,000 credits/monthNot ratedVisit
n8n CloudChecked
See sitePrices being checkedFree trial only (self-hosting is free)Not ratedVisit
ZapierChecked
$19.99/moProfessional (lowest task tier, 750 tasks)750 tasks/month at the lowest tier (page data: 'Pro 750'); multi-step Zaps; unlimited premium apps; webhooks; AI fields; email support (live chat from the 2,000-task tier)100 tasks/month, two-step ZapsNot ratedVisit

An AI agent that handles 1,000 requests a month costs the least on Make or n8n Cloud. In our worked example it fits inside Make Core ($12 a month billed yearly) or n8n Cloud Starter (€20 a month billed yearly), with the AI model paid on top or, on Make, in credits. On Zapier the same agent needs a $89 to $189 Professional plan (billed yearly), depending on the model tier, and Microsoft Copilot Studio comes to about $120 pay-as-you-go. Prices as of October 2026.

We haven’t built agents on these platforms ourselves. Zapier, Make and n8n plan prices come from AskGPU’s product data (checked October 17-18, 2026). Billing rules, Zapier’s higher task tiers and the other builders’ prices come from each vendor’s own pages, read October 27, 2026. Model prices come from our AI model data (checked October 25, 2026). The token counts in the example are illustrative, not measured.

What do no-code AI agent builders cost per month?

Each builder sells a monthly allowance in its own unit: tasks or activities on Zapier, executions on n8n, credits on the rest. That unit decides what an agent costs, more than the headline price does.

Builder Cheapest paid plan Included each month How an agent uses it Free option
Zapier (Zaps with AI steps) Professional: $19.99/mo billed yearly, $29.99 monthly 750 tasks 1, 3 or 5 tasks per AI step, plus the same per tool call 100 tasks
Zapier Agents (add-on) Pro: “~$33.33/mo” billed yearly 1,500 activities 1 per trigger, knowledge lookup, action or web search 400 activities
n8n Cloud Starter: €20/mo billed yearly, €24 monthly 2,500 executions 1 per workflow run or agent turn Trial; self-hosting free
Make Core: $12/mo billed yearly, $16 monthly (US prices) 10,000 credits 1 per module operation, plus token credits with Make’s AI 1,000 credits
Activepieces Plus: $20/mo 10,000 credits 1 per run, 2-20 per AI step, 1 per agentic action 1,000 credits
Microsoft Copilot Studio $200/mo per 25,000-credit pack, or $0.01 per credit pay-as-you-go 25,000 per pack 2 per generative answer, 5 per agent action See below
Lindy Team: $49.99/mo 5,000 credits Varies by model and task $50 of credits for 7 days

n8n prices its Cloud plans in euros only. Relevance AI’s pricing page lists only an Enterprise plan sold through sales, so it isn’t priced here.

How does each builder bill an agent?

Zapier: activities now, tasks next

Zapier Agents is an add-on: a separate subscription that Zapier’s help center says appears on your invoice apart from your Zapier plan. It counts activities, not tasks. One activity is one trigger, knowledge-source answer, action, web search, web browse or Chrome extension message. A run is capped at 10 activities on Free and 40 on paid plans, and the agent asks before going past the cap. Tests count only on Free.

Zapier is folding Agents into AI by Zapier, an AI step in the Zap editor billed in tasks. Its migration article says no date has been set to turn off Agents.

AI by Zapier steps use this formula: tasks per run = (1 × model rate) + (number of tool calls × model rate). The rate is 1 for Standard models (which can’t use tools), 3 for Advanced, 5 for Premium, and 1 with your own AI account. Zapier’s example: a Premium step with two tool calls uses 15 tasks. New AI steps on paid plans default to Premium, and a run that reaches 75 tasks pauses for approval (adjustable up to 500). Trigger steps never use tasks.

n8n: one execution per agent turn

n8n’s pricing FAQ defines an execution as one run of the entire workflow, however many steps it has. n8n launched Agents, still in preview, on September 25, 2026: one turn with an agent is one execution, and tool calls to your workflows and sub-agents don’t count separately (our report on n8n Agents). Building an agent with n8n Assistant uses Assistant credits; Starter includes 1,600 a month.

Make: credits per module, plus tokens

By default, one Make module operation uses one credit. For Make AI Agent (New), Make lists 1 credit per operation to run an agent, plus 1 per operation for each tool it calls. With Make’s own AI Provider, credits for AI tokens come on top; with your own OpenAI or Anthropic connection (paid plans only), you pay the provider instead. The feature is in open beta, and Make says its pricing may change.

Activepieces, Copilot Studio and Lindy

  • Activepieces: a flow run costs 1 credit however many steps it has. AI steps add 2 credits on a fast model, 10 on a smart model and 20 on a frontier model, and each agentic action costs 1 credit. On Plus and up, your own AI keys drop every AI step to 1 credit. The pricing page doesn’t say whether each model call inside an agent is a separate AI step; Activepieces’ blog prices one run with a frontier step at 21 credits.
  • Microsoft Copilot Studio charges 1 credit for a classic (pre-written) answer, 2 for a generative answer, 5 for an agent action (which includes triggers) and 10 for grounding in your Microsoft 365 data. Reasoning models add 10 credits per 1,000 tokens.
  • Lindy says credits cost $0.01 each on base plans, but publishes only ranges per task (2 to 250 credits for “Everyday Asks”), so an agent can’t be priced in advance.

Who pays for the AI model?

Builder Model cost included in the plan? Your own API key
Zapier (AI by Zapier) Yes, in the 1x, 3x or 5x task rate Yes; counts at 1x and your provider bills the tokens
Zapier Agents Zapier’s pricing and usage pages don’t say Not covered on those pages
n8n Cloud No Yes, or prepaid Gateway credits on Starter and Pro
Make Yes with Make’s AI Provider, as extra credits Yes, on paid plans
Activepieces Yes, in the 2, 10 or 20 credit rate Yes, on Plus and up (1 credit per AI step)
Copilot Studio Yes, for the models Copilot Studio provides Bring-your-own models are billed separately

Make publishes how many tokens one credit buys. On its Large tier (GPT-5 mini), one credit covers 3,616 input tokens or 452 output tokens; on its Small and Medium tiers (GPT-5 nano), 18,080 input or 2,260 output tokens.

n8n’s Gateway credits let Cloud Starter and Pro users run Anthropic, OpenAI, Google Gemini and other models without their own key, billed per request at the rates on n8n’s service pricing page. That page didn’t load for us, so we can’t compare the rates. Gateway credits aren’t available on self-hosted n8n.

What does an agent handling 1,000 requests a month cost?

The setup: a support agent answers 1,000 customer emails a month. For each email it is triggered by the new message, looks up the answer in a help document, and sends one reply. We assume 8,000 input and 800 output tokens per email across the agent’s model calls, an illustrative figure. We count the trigger once per email; a trigger that checks on a schedule uses credits on Make at every check (see What to watch for).

The model bill, if you bring your own key. The month comes to 8 million input and 0.8 million output tokens:

  • Claude Sonnet 5.5 ($2 input / $10 output per million tokens): 8 × $2 + 0.8 × $10 = $16 + $8 = $24.00
  • Claude Haiku 5.5 ($0.10 / $0.50): 8 × $0.10 + 0.8 × $0.50 = $0.80 + $0.40 = $1.20

Usage per email, and per month:

  • Zapier AI step with two tool calls (lookup and reply): Premium is 5 + 2 × 5 = 15 tasks, Advanced 3 + 2 × 3 = 9, your own AI account 1 + 2 × 1 = 3. For 1,000 emails: 15,000, 9,000 or 3,000 tasks. The trigger is free.
  • Zapier Agents: trigger + lookup + reply = 3 activities, or 3,000 a month. Pro’s 1,500 cover about 1,500 ÷ 3 = 500 emails; Zapier lists no larger self-serve allowance.
  • n8n: 1 execution per email: 1,000, within Starter’s 2,500.
  • Make with your own key: trigger + agent + lookup + reply = 4 credits, or 4,000 a month.
  • Make with Make’s AI Provider (GPT-5 mini): the same 4 credits plus 8,000 ÷ 3,616 = 2.21 for input tokens and 800 ÷ 452 = 1.77 for output, so about 7.98 credits per email, or about 7,980 a month.
  • Activepieces on a smart model: run 1 + AI step 10 + two agentic actions 2 = 13 credits, or 13,000 a month. With your own keys: 1 + 1 + 2 = 4 credits, or 4,000.
  • Copilot Studio: trigger 5 + generative answer 2 + reply action 5 = 12 credits, or 12,000 a month. Microsoft’s own order-processing example counts only the action calls, so your figure may be lower.

What that costs, as of October 2026, if the plan runs only this agent:

Builder and setup Monthly usage Plan needed Platform price per month Model bill
Make, Make’s AI Provider ~7,980 credits Core (10,000 credits) $12 billed yearly, $16 monthly In the credits
Make, your own key 4,000 credits Core $12 billed yearly, $16 monthly $24.00 (Sonnet) or $1.20 (Haiku)
n8n Cloud, your own key 1,000 executions Starter (2,500) €20 billed yearly, €24 monthly $24.00 or $1.20
Activepieces, your own keys 4,000 credits Plus (10,000) $20 $24.00 or $1.20
Activepieces, smart model 13,000 credits Plus + 3,000 extra $20 + 3,000 × $0.007 = $41 In the credits
Zapier, your own AI account 3,000 tasks Professional, 5,000 tasks $89 billed yearly, $133.50 monthly $24.00 or $1.20
Zapier, Advanced tier 9,000 tasks Professional, 10,000 tasks $129 billed yearly, $193.50 monthly In the tasks
Zapier, Premium tier (default) 15,000 tasks Professional, 20,000 tasks $189 billed yearly, $283.50 monthly In the tasks
Copilot Studio, pay-as-you-go 12,000 credits Pay-as-you-go 12,000 × $0.01 = $120 In the credits

What the numbers show:

  • Make and n8n come out cheapest because they don’t multiply the charge per tool call. On Zapier, each tool call costs the step’s full rate again.
  • On Zapier, the defaults cost the most. Moving from Premium to your own AI account with Claude Haiku 5.5 cuts the bill from $189 to $89 + $1.20 = $90.20 a month, billed yearly.
  • Reasoning models are expensive on Copilot Studio. At 10 credits per 1,000 tokens, 8,800 tokens add 8.8 × 10 = 88 credits per email if all of them count: 100 credits per email, or 100,000 × $0.01 = $1,000 a month on pay-as-you-go.
  • The allowance is shared with your other automations.

What do you get free?

  • Zapier Agents Free: 400 activities a month, about 133 emails at 3 activities each.
  • Zapier Free: 100 tasks and two-step Zaps; AI steps can only be previewed in unpublished Zaps.
  • Make Free: 1,000 credits a month, and Make AI Agent (New) works on every plan with Make’s AI Provider: roughly 125 emails at about 8 credits each. The minimum run interval is 15 minutes.
  • n8n: n8n Cloud offers a free trial only, but the self-hosted Community Edition is free; you pay for the server. See our self-host n8n guide.
  • Activepieces Free: 1,000 credits, agents included, no own keys: about 76 emails at 13 credits each.
  • Lindy: $50 of credits for 7 days, once per teammate.
  • Copilot Studio: no credits are charged when users licensed for Microsoft 365 Copilot use internal agents, within Microsoft’s fair-use limits.

What to watch for

  • Polling triggers on Make. A Make trigger module uses 1 credit per run even when it finds nothing new. Checking every 15 minutes uses 4 × 24 × 30 = 2,880 credits in a 30-day month. In our Make’s AI Provider example, that replaces 1,000 trigger credits with 2,880: 7,980 − 1,000 + 2,880 = 9,860, just under Core’s 10,000. Checking every minute uses 60 × 24 × 30 = 43,200. On Zapier, polling triggers use no tasks.
  • Running out. Zapier holds new Zap runs until your cycle resets unless you turn on pay-per-task (1.25 times the base rate on yearly plans, 2.5 times on monthly, up to 3 times your allowance). Zapier Agents stop running actions, searches and triggers. Copilot Studio disables custom agents at 125% of prepaid capacity. Activepieces Free runs wait until credits refresh.
  • Extra credits cost more. Make charges 25% more for extra credits than for plan credits, and its auto-purchase buys 10,000 at a time.
  • Unused credits disappear. Make credits expire at the end of the term, Lindy’s plan credits and Copilot Studio pack credits don’t roll over, and n8n Gateway top-ups expire after 12 months.
  • Beta pricing. Make AI Agent (New) is in open beta, n8n Agents are in preview, and Zapier is moving Agents into AI by Zapier. Billing could change.
  • Currency. n8n Cloud is priced in euros, and Make’s prices depend on your country.
  • Long conversations. Lindy’s docs say every message in a thread re-reads everything above it, so long threads cost more per turn.

Which should you choose?

  • Lowest bill for a steady email or chat agent: Make. Core at $12 a month billed yearly covered our 1,000 emails even with Make’s own AI, as long as the trigger doesn’t poll often.
  • Many tool calls per request, or a technical team: n8n, where a turn is one execution however many tools it calls. Self-hosting removes the execution cap; see self-hosted AI agents.
  • Already built on Zapier: use AI by Zapier steps with your own AI account or the Advanced tier, not the Premium default. Agents Pro fits about 500 such requests a month.
  • Open source with flat pricing: Activepieces Plus at $20 a month, with your own model keys.
  • Microsoft 365 companies building agents for staff: Copilot Studio, especially where users hold Microsoft 365 Copilot licenses.
  • Lindy: only after its free week shows what your tasks really use.

For more automation tools and their plans, see our automation hub.

Sources