Previous lesson
Benchmark score vs production cost
Continue the Model Costs track.
← Back to Compute College
Compute College
How to estimate cost per completed AI task
Learn how to estimate the full cost of an AI task, including input tokens, output tokens, retries, tool calls, latency, and model selection.
Compute & Pricing LessonsLearning path
One concept connected to AI compute market decisions.
5-8 minutesRead time
A practical introduction designed to be completed in one sitting.
Task Cost / Agents / InferenceTags
Useful for developers, founders, procurement teams, and analysts tracking model-serving economics.
Plain-English definition
Plain-English definition
Cost per completed AI task is the total model-serving spend required to finish one useful workflow, including failed attempts and intermediate calls, rather than only the price of the first prompt.
Why it matters
Why it matters
This unit turns provider prices into buyer math. Coding agents, research agents, and verification workflows often invoke a model several times before one result is usable.
- Capability changes matter economically only when they affect deployed workloads or buyer choices.
- Token volume, latency, retries, and throughput determine how a useful result becomes serving cost.
- A ComputeTape reader should connect model evidence to inference demand and required AI compute capacity.
Simple example
Simple example
An illustrative task uses 20,000 input tokens at $5 per million and 4,000 output tokens at $25 per million: $0.10 plus $0.10, or $0.20 per attempt. If only four of five attempts succeed, expected cost per completed task is $0.20 / 0.80 = $0.25 before tool fees or overhead.
- Use the example to compare workload economics, not as a current market quote.
- Record the task type, evaluation or workload conditions, and the cost inputs before comparing results.
- A successful result is valuable only if its latency and cost fit the intended production use.
Example figures are illustrative calculations, not current quoted market prices.
Market signal
How to read the market signal
A model may increase demand while lowering cost per successful task: higher completion rates can make many more workflows economically viable at the same posted token rates.
- Look for adoption, routing, usage-volume, or capacity signals rather than a headline score alone.
- Compare input tokens, output tokens, latency, tool rounds, retries, and completion quality together.
- Keep sourced capability facts separate from interpretation about future AI compute demand.
Market read: this metric becomes an AI compute signal only when it changes serving volume, effective workload cost, or the capacity buyers require.
Common mistake
Common mistake
Do not price only the first response when the real workflow routinely includes retries, verification, context refreshes, and tool rounds.
Practical takeaway
What you can do with this
Estimate input cost plus output cost plus tool or orchestration cost for all expected attempts, then divide by the probability of an acceptable completion.
- Buyers: test the metric on tasks close to the workload you will pay to serve.
- Builders: measure tokens, latency, retries, completion rate, and model price on each test run.
- Analysts: require a source and an adoption mechanism before treating a model result as demand evidence.
Decision check: identify the capability measured, the serving cost driver it affects, and the buyer behavior that would make capacity demand change.
Helpful memory trick
Helpful memory trick
The useful unit is not cost per prompt. It is cost per finished job.
Compute College
Follow model releases as market signals
Follow model releases as AI compute market signals in the ComputeTape Morning Brief.
Compute College track
Model Costs
Continue this Compute College lesson path
Next lesson
Model latency explained
Continue the Model Costs track.