1,500 < 1,600
Monica Pro’s entire monthly credit allowance is 1,500. One Veo 3 video costs 1,600. The plan cannot afford a single video generation in a month, and nothing on the pricing page tells you that — because there is no pricing page text.
Credit decoder · By Adam Johnstone · Updated 2026-09-13
What does one credit actually buy?
Queries, credits, points, tokens — every bundle invents a currency and sets its own exchange rate to real work. This page converts each one back into messages, images and documents, using only figures the vendors publish themselves. Where a vendor publishes too little to compute, the cell says so.
The decoder
What a month's allowance actually buys.
Every bundle meters in its own currency. This converts each one into real work, using the vendor's own published figures — and marks the cells where a vendor publishes too little to compute anything.
| Plan | Currency | Monthly allowance | Published cost per action | What that actually buys |
|---|---|---|---|---|
| ChatHub Pro | Queries, split basic / advanced | 10,000 basic + 1,500 advanced | Not published per action | 50 advanced messages a day, or 333 basic. Image credits (3,000/mo) cannot be converted — ChatHub publishes no per-generation cost. |
| Monica Pro | Advanced credits + query count | 200 advanced queries + 1,500 credits | GPT-4o 2 credits / 1k tokens · DALL·E 3 30 / image · Veo 3 1,600 / video | 7 advanced queries a day. The 1,500 credits buy 50 DALL·E 3 images — or not one Veo 3 video, which costs 1,600. Published costs cover retired models only. |
| Sider | Unified credit pool | Not published | Deep Research 20 · Scholar 10 · ChatPDF 1–28 · translate 1–15 · audio 1 per 10 min | Uncomputable. Sider publishes the most detailed per-action costs in the category and never states how many credits a plan includes. |
| Merlin Free | Queries | 102 per day | Claude 3 Opus 50 · GPT-4o 15 · Flux 1 Pro 140 / image · web access doubles cost | 2 Claude-Opus-class messages a day, or 6 GPT-4o. One premium image costs 140 — more than a full day’s allowance. Costs published for retired models only. |
| MultipleChat Pro | Messages + Studio tokens | ~120 messages/day + 1M Studio tokens | Not published | Roughly 120 messages a day, vendor’s own word “about”. What 1M Studio tokens produces is not stated — no per-document cost is published. |
| TypingMind | None — direct API billing | Unlimited | Your provider’s published token prices | Whatever you pay for. The only tool here where cost per task is knowable in advance, because the rate card is the provider’s and it is public. |
Checked 13 September 2026. Where a cell says uncomputable, the vendor publishes an allowance without per-action costs, or per-action costs without an allowance. No figure here is estimated or borrowed from a third party.
Three numbers worth sitting with
What falls out of the conversion.
140 > 102
Merlin’s free tier gives 102 queries a day. One Flux 1 Pro image costs 140 queries. The headline number sounds generous until you convert it into the thing you wanted to do.
50× inside one plan
On Merlin’s published table, a Claude 3 Opus message costs 50 queries and a GPT-4o Mini message costs 1. Same subscription, same allowance, fiftyfold difference. “102 queries a day” means between 2 and 102 messages.
Why this page exists
The complaint nobody answers.
Credit opacity, not price, is the recurring complaint about this category. Public reviews of credit-based AI tools describe allowances vanishing without explanation, undocumented exchange rates, and users asking vendors for a cost estimate before running a task rather than discovering it after.
The structural cause is visible in the table above: vendors publish half the equation. An allowance with no per-action cost is a number you cannot budget against. Per-action costs with no allowance is a price list for a wallet of unknown size. Two of the six plans here publish both halves, and one of those two is the tool that has no credit system at all.
Three of the six publish cost tables that price only models the vendor has retired. Those tables are not wrong so much as abandoned — and an abandoned cost table is worse than none, because it looks like disclosure.
FAQ
Questions about credits.
How do AI bundle credits actually work?
Every bundle meters usage in its own invented currency — queries, credits, points, tokens — and the exchange rate to real work is set by the vendor and changeable at will. Merlin's own cost page states that the team “re-adjusts query mapping on a dynamic basis”. To know what a plan buys you need two numbers: the monthly allowance and the cost per action. Most vendors publish one and not the other.
How many messages does an AI bundle credit buy?
It depends on the model, and the spread is enormous. On Merlin's published table a Claude 3 Opus message costs 50 queries and a GPT-4o Mini message costs 1 — a fiftyfold difference inside one subscription. On Monica, credit burn ranges from roughly 0.2 credits per 1,000 tokens on budget models to 40 on reasoning models. An allowance quoted without a model is not information.
Which AI bundle has the clearest credit system?
TypingMind, because it has none — you pay your provider directly at published token rates, so cost per task is knowable before you run it. Among credit systems, ChatHub publishes allowances clearly but no per-action costs; Sider publishes per-action costs in detail but no allowances. Neither gives you both halves.
Why can't you calculate what Sider or MultipleChat credits buy?
Because each publishes only half the equation. Sider details what every action costs — 20 credits for a Deep Research topic, 1 to 28 for a ChatPDF session — but never states how many credits Lite, Pro or Max include. MultipleChat gives a monthly Studio token figure but no per-document cost. Without both numbers no honest estimate is possible, so this page leaves those cells empty rather than guessing.
Do image and video generation use the same credits as chat?
Usually yes, out of one pool, and they are dramatically more expensive. One Flux 1 Pro image on Merlin costs 140 queries against a free daily allowance of 102. One Veo 3 video on Monica costs 1,600 credits against a Pro plan's entire monthly allowance of 1,500. A single generation can consume more than a month of chat.
Do unused AI credits roll over?
Rarely, and the reset date varies in ways that cost you money. Monica's credits reset on the first of the calendar month, so signing up on the 20th gives you a partial allowance for full price. Sider resets on your billing anniversary instead, and its booster-pack credits never expire while monthly credits do. Check which model a vendor uses before you buy mid-month.