What is artificial intelligence cost per completed task?
Definition
Artificial intelligence cost per completed task measures economic efficiency at the outcome level. Instead of looking only at the listed price per token or request, it totals the model usage required for a task that meets a defined quality threshold.
The metric can include failed attempts, revisions, tool calls and human correction. It helps compare models with different token efficiency and reliability. A higher-priced model may have a lower completed-task cost if it succeeds quickly, while a cheap model may remain preferable for predictable routine work.
Acronyms and aliases
cost per completed task synonymcompleted-task cost variant
Related terms
Frequently asked questions
Why is completed-task cost better than token price alone?
Token price ignores whether the output solved the problem. Completed-task cost accounts for the requests and retries needed to reach an acceptable result.
What should be included in AI task cost?
At minimum include model usage for successful and failed attempts. A fuller calculation may add tool charges, latency, infrastructure and human review time.
Videos explaining artificial intelligence cost per completed task
