AI Cost per Outcome in Production, by Workload

Cost per successful outcome ranges from a fraction of a cent to roughly sixteen cents. The low end is a single RAGRetrieval-Augmented Generation (RAG) retrieves relevant passages at query time and feeds them into an LLM's context. Q&A call on a small model. The high end is a multi-step agentic tool loop on a frontier model. We modeled four common workloads. The gap between a clean-call sticker price and an actual successful outcome runs from about 1.2x to more than 3x. Retries, growing context, and failed attempts drive that gap. These are modeled numbers, not measured client telemetry.

Every price, token count, and multiplier below is stated, so you can rederive each number yourself. This piece is the concrete counterpart to what an AI feature costs in production. That piece explains why the gap exists. This one puts numbers on it.

How we modeled cost per successful outcome

The unit that matters is cost per successful outcome, not cost per call. A call that fails still costs money, and something else has to pay for it. The formula:

cost_per_successful_outcome = (input_tokens × input_price + output_tokens × output_price) × overhead_multiplier ÷ success_rate

input_tokens × input_price + output_tokens × output_price is the clean-call sticker price: what one attempt costs if it works the first time. The overhead_multiplier folds in retries, fallback double-billing, and the extra tokens a growing context re-sends on every turn. Dividing by success_rate turns a per-call price into a per-outcome price. If four out of five attempts succeed, the successful ones have to absorb the cost of the one that did not.

Prices (public list prices as of 2026-07-11; representative tiers, not a specific vendor SKU, since list prices move):

• A small/fast model tier ≈ $0.15 / $0.60 per 1M input/output tokens.
• A frontier model tier ≈ $3 / $15 per 1M input/output tokens.

Per-workload assumptions used to compute every cell below:

• RAG Q&A: one call, ~4,000 input tokens (retrieved chunks + query) plus ~400 output tokens; overhead 1.1x; success rate 95%.
• Document extraction: one call, ~8,000 input tokens (one document) plus ~800 output tokens (structured JSON). Overhead is 1.25x for reparse and repair passes; success rate 90%.
• Chat with history: one 10-turn session modeled as a single outcome. Input grows linearly from ~500 to ~5,000 tokens per turn (27,500 tokens total), plus 500 output tokens per turn (5,000 total). Overhead is 1.15x; success rate 92% for the whole session.
• Agentic tool loop: five chained model calls per outcome, ~2,000 input and ~300 output tokens each (10,000 in / 1,500 out total). Context grows across the chain. Overhead is 1.4x. Success rate is 85% per step. That compounds to roughly 44% for the whole five-call outcome (0.85⁵).

What each workload costs

Prices as of 2026-07-11; modeled from the assumptions above, not measured client telemetry.

A single RAG answer on a small model is cheap either way. It costs a tenth of a cent whether you count the clean call or the successful outcome. Document extraction and chat sessions cost more mainly because they push more tokens, not because the multiplier is dramatic. The agentic tool loop is the outlier. Its modeled cost is more than three times its sticker price, because chaining five calls means chaining five chances to fail.

Why the modeled cost runs higher than the sticker price

The gap between a clean-call sticker price and a modeled cost per successful outcome comes from two places. They do not contribute equally.

The overhead multiplier (1.1x to 1.4x here) covers retries on malformed or rate-limited calls. It also covers a fallback provider billed on top of the original attempt. It covers the extra tokens a growing context re-sends every turn too. It is a real cost. On its own, it is a modest tax: 10% to 40% in this model.

The success-rate divisor is where multi-step workloads get expensive, because failure compounds. RAG Q&A and document extraction are single calls, so their stated success rate applies once. The agentic tool loop chains five calls. If each step succeeds 85% of the time independently, the chain as a whole succeeds roughly 0.85⁵, or 44% of the time. More than half of all attempts fail somewhere along the way. The outcomes that do succeed have to cover the cost of all of them. That is why the agentic loop’s modeled cost runs about 3.2x its sticker price while RAG Q&A’s runs about 1.2x. Its per-step overhead and success numbers do not look dramatically worse alone. Cost adds with each chained call. Failure multiplies.

This is the same mechanism the qualitative piece on AI cost describes: retries, fallbacks, and growing context billing twice. Here it is modeled with actual numbers instead of a general warning. The wider 5x-20x gaps teams report in the wild usually stack a second effect on top of this one. Production prompts run far longer than whatever they benchmarked against the pricing page.

What this means for a feature you’re about to ship

Before you pick a model tier, model the workload shape, not just the per-token price. A frontier model looks like a 20x price jump over a small/fast one on the pricing page. For a single RAG call, that is the difference between $0.00097 and $0.0208 per outcome. Both are trivial at almost any volume. What matters is how many chained calls an outcome needs, and how reliable each one is. That is what turns a modest overhead multiplier into a 3x cost multiplier. A tool loop can be cheap to call once. But it calls itself five times per outcome at 85% reliability per step. That costs more than its sticker price suggests, by a margin wider than switching model tiers will ever close.

Run your own workload’s numbers through the same formula. Plug in your token counts, tier, and expected success rate at the cost calculator. Do it before you commit to an architecture, not after the first month of invoices.

For the mechanisms behind a real bill running ahead of the modeled one, see two field guide entries: retry-amplified spend and prompt cache never hits.

Once the numbers say where your spend is going, the AI cost model shows the levers that move it. Six surfaces, and what each one costs you to pull.