← Retour au plan
Credits & Pricing on OUPI: The Complete Beginner's Guide · Leçon 8 sur 8

8. Tips to Keep Costs Low

Apply practical strategies — automatic model selection, clear briefs, knowledge bases over pasting, and temporary chats — to minimize credit consumption.

Every action on OUPI — a chat message, an image generation, an agent mission — costs credits based on the resources it actually consumes. Text costs depend on the number of tokens exchanged multiplied by the weight of the model you use: frontier (top-tier) models cost far more per token than fast (lightweight) ones. Media is priced per generation depending on engine and output size or duration. Agent missions add up all model calls and tool uses across their work rounds. The good news: you are never charged for an operation that fails, and costs are always shown so you can learn what is expensive.

What Drives Credit Cost💬 Text ChatsTokens × Model WeightFrontier = $$$Fast = $🖼️ MediaEngine + Size/DurationPer generation🤖 Agent MissionsAll calls + toolsSummed across rounds✓ Failed operations are never charged
Cost factors: what drives credit consumption
Astuce

Let automatic model selection do the work. When you don't manually pick a model, OUPI favors lighter, cheaper models for routine tasks and only escalates to heavier ones when the complexity demands it. For most everyday questions this saves credits with no noticeable quality loss.

Astuce

Write a clear, specific brief for every mission. Vague goals force the agent to run extra rounds of reasoning and tool use, each of which costs credits. A well-defined outcome means fewer rounds and a lower bill. As the platform itself says: a clear brief is the cheapest optimization.

Astuce

Store long or reusable documents in a knowledge base instead of pasting them into the chat. Pasted text inflates the token count of every message in the conversation, while a knowledge base lets the model retrieve only the relevant parts — dramatically cutting token usage.

Pasting vs. Knowledge Base❌ Pasting Full DocumentMsg 1: doc + question (5000 tok)Msg 2: doc resent (5000+ tok)Msg 3: doc resent again…Tokens snowball 📈✅ Knowledge BaseDoc stored once externallyOnly relevant parts retrievedTokens stay low ✓
Pasting documents vs. using a knowledge base
Astuce

Use temporary chats for throwaway questions. A persistent conversation accumulates context (and therefore tokens) over time. If you only need a quick answer, a temporary chat keeps the token count minimal and avoids bloating future messages.

4 Habits for Low Costs1Auto ModelSelection2Clear, SpecificBriefs3Knowledge Baseover Pasting4TemporaryChats= Fewer tokens, fewer rounds,lower credit spend 💰
Four habits to minimize credit consumption
À vous de jouer

Open the Missions page and look at the cost shown on one of your recent missions. Compare a mission where you gave a detailed brief with one where the goal was vague. Notice the difference in rounds and total credits consumed — this builds your intuition for what is expensive.

Suivre ce cours dans OUPI → Cet exercice se fait dans la plateforme OUPI.
À retenir

Four habits keep your credit spend low: (1) let automatic selection choose the lightest model that fits the task, (2) write clear, specific briefs so missions finish in fewer rounds, (3) use knowledge bases instead of pasting long documents, and (4) reach for temporary chats when you don't need a lasting conversation. Check mission costs regularly to sharpen your sense of what consumes the most — and remember, you're never charged for a failed operation.