unitcostai

Compare Model Costs

What happens to cost and margin if I switch models?

How this tool works

This tool runs the same deterministic workflow economics model twice (baseline and candidate), then reports cost and margin deltas so you can decide if an agent or assistant-model migration is financially justified.

How It Works

  1. Select baseline and candidate model rows from the pricing snapshot.
  2. Keep usage assumptions identical across both runs.
  3. Compare cost per user, gross margin, and break-even price deltas.

Formula

margin_delta_pct = margin_candidate_pct - margin_baseline_pct

cost_delta_per_user = cost_candidate - cost_baseline

Assumptions and Units

  • Currency: USD
  • Token unit: token
  • Both runs share the same usage and retrieval assumptions
  • Pricing source: daily pricing snapshot in repo, no runtime scraping

Related resources

AI Workflow Cost, Agent Run Cost, Tool Use Cost, Human-in-the-Loop Cost, Break-even Price, AI Cost Estimation, AI Coding Agent Cost per Developer, AI Analyst Workflow Cost per Analyst, Model Selection: Quality vs Unit Cost, How Model Switching Changes Margin.

Pricing snapshot: 2026-07-29Baseline: OpenAI/GPT-5.6 SolCandidate: Anthropic/Claude Fable 5

Step 1 Baseline and Candidate Models

iSet the current model as baseline and the proposed model as candidate.

Step 2 Quick Mode

iSet shared usage and pricing assumptions before tuning advanced variables.

Steady usage, moderate context, and balanced quality-cost tradeoffs.

iSuggested defaults are transparent and editable.

Optional Advanced assumptions

iTune retrieval, reranking, embeddings, vector, caching, and infra.
Show advanced inputs

Scenario actions

Copy scenario URL

Paste into ChatGPT or Claude, or share with a teammate.

Save and track this scenario

Track pricing drift on this scenario and get an email if the latest result changes.

How tracking works

After you click Save and track, we carry this exact calculator state into the tracked-scenarios page so you can sign in and confirm the save.

We save your assumptions and the pricing snapshot used for this result.

When a newer pricing snapshot lands, we recompute the same scenario, show what changed, and email you if the latest result moved.

1 tracked scenario free, then $12/mo or $120/yr for up to 25 tracked scenarios.

Decision Signal

Candidate model reduces margin

Switching to Claude Fable 5 changes gross margin by -2.6% and cost per user by +$1.0017.

Baseline cost / user / month

$2.3634

Candidate cost / user / month

$3.3651

Margin delta

-2.6%

Monthly cost impact

+$651.11

Top Cost Drivers

iLargest baseline sensitivity shifts when each variable is increased by 10%.
Requests Per User Month10.0%
Rerank Docs4.8%
Cache Hit Rate-4.3%

Totals

iBaseline vs candidate totals and deltas under the same usage assumptions.

Monthly gross profit impact uses 650 active users.

Cost per request
Baseline
$0.02626
Candidate
$0.03739
Delta
+$0.01113
Cost per user/month
Baseline
$2.3634
Candidate
$3.3651
Delta
+$1.0017
Gross margin %
Baseline
93.9%
Candidate
91.4%
Delta
-2.6%
Break-even price
Baseline
$2.3634
Candidate
$3.3651
Delta
+$1.0017
Monthly gross profit impact
Delta
-$651.11
MetricBaselineCandidateDelta
Cost per request$0.02626$0.03739+$0.01113
Cost per user/month$2.3634$3.3651+$1.0017
Gross margin %93.9%91.4%-2.6%
Break-even price$2.3634$3.3651+$1.0017
Monthly gross profit impact-$651.11

Component Breakdown (USD/user/month)

iEach component is computed independently then summed for both models.
GenerationiModel input/output token spend for requests.
Baseline
$1.269
Candidate
$2.25
Delta
+$0.981
RetrievaliExtra model input spend from retrieved context chunks.
Baseline
$0.45
Candidate
$0.9
Delta
+$0.45
RerankingiReranker cost based on docs scored per request.
Baseline
$1.62
Candidate
$1.62
Delta
$0
Embeddings IngestioniAmortized per-user share of the fixed monthly corpus embedding refresh cost.
Baseline
$0
Candidate
$0
Delta
$0
Vector DbiVector database query cost across all requests.
Baseline
$0.0014
Candidate
$0.0014
Delta
$0
CacheiSavings from cache hits. Negative means lower total cost.
Baseline
$-1.0129
Candidate
$-1.4422
Delta
-$0.4293
InfraiNon-model infra overhead per request.
Baseline
$0.036
Candidate
$0.036
Delta
$0
ComponentBaselineCandidateDelta
GenerationiModel input/output token spend for requests.$1.269$2.25+$0.981
RetrievaliExtra model input spend from retrieved context chunks.$0.45$0.9+$0.45
RerankingiReranker cost based on docs scored per request.$1.62$1.62$0
Embeddings IngestioniAmortized per-user share of the fixed monthly corpus embedding refresh cost.$0$0$0
Vector DbiVector database query cost across all requests.$0.0014$0.0014$0
CacheiSavings from cache hits. Negative means lower total cost.$-1.0129$-1.4422-$0.4293
InfraiNon-model infra overhead per request.$0.036$0.036$0
Sensitivity RankingiBaseline sensitivity: cost change when one variable is increased by 10%.
VariableDelta cost %
Requests Per User MonthiUser activity level per month.10.0%
Rerank DocsiDocs reranked per request.4.8%
Cache Hit RateiFraction of requests served by cache.-4.3%
Output TokensiGenerated tokens per request.2.6%
Retrieved ChunksiRetrieved chunk count per request.1.3%
Tokens Per ChunkiAverage chunk size in tokens.1.3%
Input TokensiPrompt-side tokens per request.1.2%
Vector Queries Per RequestiVector query count per request.0.0%
Monthly Active UsersiActive-user estimate used to amortize fixed monthly embedding refresh.-0.0%

Assumptions and Units

  • CurrencyUSD
  • Token unittoken
  • Pricing snapshot2026-07-29
  • Baseline modelOpenAI/GPT-5.6 Sol
  • Candidate modelAnthropic/Claude Fable 5
  • Comparison ruleUsage, retrieval, and fixed monthly terms stay shared across both runs

Recommended Next Step

iUse this section to turn your model switch results into the next infrastructure or routing checks.

If retrieval or hosting is driving cost, check infra options first, then come back to re-check the switch math.

Model and Routing Preparation

Model Selection: Quality vs Unit CostAI Workflow Cost

Compare infra providers

View Infra Recommendations

Sources and Snapshot

iPricing comes from the current dated snapshot.

Active Pricing Rows

Baseline

OpenAI / GPT-5.6 Sol

  • Input tokens$5 / 1M
  • Output tokens$30 / 1M

Candidate

Anthropic / Claude Fable 5

  • Input tokens$10 / 1M
  • Output tokens$50 / 1M

Shared retrieval defaults

  • Embedding input$0.02 / 1M
  • Rerank docs$1 / 1K

Continue Analysis

Switch tools

Read guides