How are credits consumed on Merlio?

Credits are consumed per message, based on the AI model used and the size of the request. Short questions cost less than long ones, and premium models like Claude Opus or GPT-5.4 Pro cost more than smaller models like GPT-5-nano or Haiku.

Why does the same model show different credit costs?

The cost shown next to a model is the average per use. Actual cost varies with how much you write (input) and how much the AI replies (output). Saying 'Hi' might cost 1 credit, while asking for a 200-line code review on the same model can cost 50+ credits.

Are credit costs the same for image generation?

Image generation uses a fixed cost per image, depending on the model (e.g. Flux Schnell, Imagen 4, GPT Image 2). Higher-quality models cost more credits per generation.

Do unused reserved credits get refunded?

Yes. Merlio reserves a maximum amount when you start a generation, and refunds the unused portion automatically once the AI has finished responding. You only pay for what you actually use.

AI Chat Image Generator Interior Design Exterior Design Tattoo Generator Food Photography Video Generator Prompt Builder

All Tools AI Models Pricing Blog Affiliate

Credit Usage Policy

How credits are consumed on Merlio

Name: Merlio
Rating: 4.5 (127 reviews)
Author: Merlio

Every AI message has a credit cost. The exact amount depends on two things: the model you use and the size of your request.

Per model

Premium models (Claude Opus, GPT-5.4) cost much more per token than smaller ones (GPT-5-nano, Haiku 4.5). Each model page shows its average credit cost upfront.

Per message size

AI bills both your input (the prompt) and output (the reply). "Hi" costs almost nothing. A 200-line code review, long conversation, or detailed essay costs more.

Image generation

Image models charge a fixed cost per image. Model choice (Flux, Imagen, GPT Image 2) determines the price.

Video generation

Video models charge a cost that scales with duration. A 3-second clip is much cheaper than an 8-second one. Range shown per model below.

Real example: same model, very different costs

Say you're using GPT-5.4 (input $2.50 / output $15 per 1M tokens). Here's what two different messages cost:

Tiny message

You: "Hi"
AI: "Hello! How can I help?"

~10 tokens total → about 1 credit

Heavy request

You: "Review this 200-line Python script…"
AI: <detailed line-by-line review, refactor suggestions>

~3 000 tokens total → about 50–60 credits

Reservation & refund: When you start a generation, Merlio reserves the maximum possible cost (worst case). Once the AI finishes, the unused portion is refunded automatically, you only pay for what was actually consumed.

The exact average cost per use, for every model on Merlio, is listed below. Filter, search, and sort to find the model that fits your budget.

267

Total models

122

Chat models

142

Image models

Video models

Free models

Showing 1–20 of 267

Chat models show ~average cost per message — varies with prompt size and reply length.Image models show fixed cost per generated image.Video models show a range — final cost depends on the video duration you pick.