Cost = (input tokens ÷ 1M × input price) + (output tokens ÷ 1M × output price). Prices come from our daily-verified model data. Batch and cached-input discounts are not applied — they only ever make these numbers smaller.
“Needs chunking” means the model's context window can't hold this job in a single pass, so the real cost is higher than the figure shown and quality usually suffers.
Should you just pick the cheapest?
A whole job search costs pennies at API rates, so almost nobody should be paying per token for this — the free consumer tiers cover it comfortably. The API figures below matter only if you are building a tool that writes letters for other people.
How much does it cost to write a cover letter with AI?
$0.0001 on Mistral: Mistral Nemo, the cheapest capable option, rising to $0.0050 on Claude Sonnet 5 at the top end. GPT-5.6 Luna is the value pick at $0.0006 per run. The job is priced at 2,200 input and 550 output tokens — see the working below.
How did you work out the token count for this task?
A two-page résumé plus a full job posting ≈ 1,650 words ≈ 2,200 input tokens. A 400-word letter ≈ 550 output tokens.
What does this cost at 25 applications a month?
$0.0015 a month on Mistral: Mistral Nemo, $0.014 on GPT-5.6 Luna, and $0.124 on Claude Sonnet 5. Batch APIs typically halve these figures for work that can wait, and prompt caching cuts the input side further when the same context is reused.
Is the cheapest model the right choice for this task?
A whole job search costs pennies at API rates, so almost nobody should be paying per token for this — the free consumer tiers cover it comfortably. The API figures below matter only if you are building a tool that writes letters for other people.
Are these prices current?
Yes. Every figure on this page is computed from our model pricing data, which is checked daily against each provider's official pricing page. When a provider changes a price, these numbers change with it.