Cost guide

- The plugin itself is free. No subscriptions, no markup, no upsell paywall. - You pay your AI provider directly for the API calls the plugin makes. - Most blogs spend under $5/month. With GPT-4o Mini you can generate ~10,000 replies for…

TL;DR

  • The plugin itself is free. No subscriptions, no markup, no upsell paywall.
  • You pay your AI provider directly for the API calls the plugin makes.
  • Most blogs spend under $5/month. With GPT-4o Mini you can generate ~10,000 replies for $1.
  • A daily cap (default 100 replies/day) prevents runaway spend. It's filterable.

What costs money

Two things — only the second one costs anything:

Item Cost
ReplyMind plugin $0 — free, GPL v2, no recurring fee
AI API calls Billed by OpenAI or Anthropic at their published rates

There's no markup, no middleman. The plugin authenticates with your API key and calls the provider directly.

Cost per reply, by model

These are approximate — actual cost depends on prompt length and reply length, but they hold for the plugin's defaults (Short replies, ~300 input tokens of context + ~100 output tokens).

OpenAI

Model Approx. cost per reply Replies per $1 Notes
gpt-4o-mini (default) ~$0.0001 ~10,000 Recommended starter
gpt-3.5-turbo ~$0.0002 ~5,000 Older, similar cost
gpt-4o ~$0.0015 ~660 Higher quality, ~10× cost
gpt-4-turbo ~$0.005 ~200 Older premium tier
gpt-4 ~$0.015 ~65 Most expensive

Anthropic Claude

Model Approx. cost per reply Replies per $1 Notes
claude-haiku-4-5-20251001 ~$0.0003 ~3,300 Fastest, cheapest
claude-sonnet-4-6 (default) ~$0.003 ~330 Recommended balance
claude-opus-4-6 ~$0.025 ~40 Most expensive

What drives the cost per reply

Three variables, in order of impact:

  1. Model choice. Switching from gpt-4 to gpt-4o-mini drops cost by ~150×.
  2. Response length setting. A Detailed reply uses 3–4× more output tokens than a Short reply.
  3. System prompt length. Every word you add to the System Prompt field adds input tokens to every reply. A 50-word prompt is fine; a 500-word prompt adds noticeable cost over thousands of replies.

What does not materially affect cost: the comment length itself (capped by wp_strip_all_tags and the post excerpt cap of 40 words), the number of providers you've signed up with, or whether the reply is in suggestion or auto mode.

Worked examples

A typical small blog

  • 50 new comments per month.
  • Suggestion mode, Short replies, GPT-4o Mini.
  • Cost: 50 × $0.0001 = $0.005 per month (effectively free).

A mid-volume site

  • 1,000 new comments per month.
  • Auto mode, Medium replies, Claude Sonnet.
  • Cost: 1,000 × $0.005 ≈ $5 per month.

High-volume / news site

  • 10,000 new comments per month.
  • Suggestion mode, Short replies, GPT-4o Mini.
  • Cost: 10,000 × $0.0001 = $1 per month.

Initial backfill on an old blog

  • One-time batch: 5,000 existing comments.
  • Suggestion mode, Short replies, GPT-4o Mini.
  • Cost: 5,000 × $0.0001 = $0.50 for the entire backfill.

How to keep costs predictable

1. Use the daily rate limit

The plugin caps at 100 successful API calls per day by default. If a spam wave breaks through approval, or you accidentally enable auto mode on a high-volume site, the cap kicks in and stops further calls until midnight.

Adjust it with a filter:

// Allow up to 500 replies per day
add_filter( 'replymind_daily_limit', fn() => 500 );

// Disable the cap entirely
add_filter( 'replymind_daily_limit', fn() => 0 );

// Different cap on weekends
add_filter( 'replymind_daily_limit', function () {
    return ( wp_date( 'N' ) >= 6 ) ? 50 : 200;
} );

Drop into your theme's functions.php or a small mu-plugin.

2. Set provider-side spending limits

Both OpenAI and Anthropic let you set a hard monthly spending cap on your account. We strongly recommend you do this:

  • OpenAI: Settings → Billing → Usage limits → Set monthly budget.
  • Anthropic: Console → Plans & Billing → Set monthly budget.

This is your second line of defence after the plugin's own daily cap.

3. Pick a cheap model first

Start with GPT-4o Mini or Claude Haiku 4.5. They're more than good enough for 90% of comment replies. Upgrade only if you see consistent quality issues.

4. Use Short or Medium response length

Detailed (2–3 paragraphs) burns 3–4× the tokens of Short. For typical reader comments, Short is the right default — most people don't expect a thesis in response to "Great post!".

5. Skip low-value comments via filter

// Skip very short comments (probably "thanks!" or spam)
add_filter( 'replymind_send_comment_data', function ( $allow, $comment ) {
    if ( str_word_count( $comment->comment_content ) < 5 ) {
        return false;
    }
    return $allow;
}, 10, 2 );

This both saves money and avoids generating throwaway replies.

Free credits to start with

Both providers offer free credits to new accounts (terms change, check current offers):

  • OpenAI: Often $5–18 of free credit on signup, expiring in 3 months.
  • Anthropic: Free trial credit on Console signup; varies.

That's enough to run tens of thousands of replies on the cheapest models — easily enough to evaluate the plugin before paying anything.

Will my AI provider charge me even if I never get any comments?

No. AI providers bill per API call, not per month. If your site gets zero new comments and you don't run any batch operations, your AI bill is zero. There are no idle/standby charges.

How to monitor your spending

Check directly with your provider:

The plugin doesn't track or report API costs in its own UI. The provider's dashboard is the source of truth.

Read next