Section 1 of 1
Understanding & Optimizing Credit Usage
creditsusagemanagementbilling
Credits fuel your AI operations. Each prompt execution, model comparison, and batch processing run consumes credits based on the model used and tokens generated.
Single prompt execution (GPT-4o)
1-3 creditsDepends on prompt + output length
Single prompt execution (Claude 3.5)
2-4 creditsHigher due to output quality
Single prompt execution (Haiku/3.5-turbo)
0.2-0.5 creditsBest for high-volume, lower-stakes tasks
Lab multi-model compare (3 models)
3-9 creditsSum of individual model costs
Batch processing (per row)
Same as singleVolume discounts on Pro/Enterprise
AI Rewrite (✨ button)
0.5 creditsUses a fast model for refinement
Pro Tip: Credit Conservation
The single biggest credit saver: Write better prompts. A precise prompt that generates the right output on the first try uses 1 credit. A vague prompt that requires 5 iterations uses 5 credits. Invest time in the Prompt Architect to save credits downstream.