Token prices for 24 current models from 8 providers (OpenAI, Anthropic, Google, xAI, DeepSeek, Mistral, Meta, Perplexity), projected against your workload. Per-seat prices for 14 tools, M365 Copilot through Cursor and Kiro, against your headcount. Filter to what you use. Runs in your browser: no signup, no tracking, and it works offline.
On iPhone: tap Share, then "Add to Home Screen" to install.
Custom rows live in your browser only.
| Model | Trend | $ in/M | $ out/M | $/day | $/month | vs cheapest |
|---|
Cache modelling: cached input billed at 10% of the input rate (vendor mechanics vary; verify for your stack). Marks on model names carry long-context, intro-pricing and retirement caveats.
Prices are the published per-user monthly rate on annual billing where both exist. The notes carry monthly rates, seat minimums, credit quotas and licence prerequisites. Quotas differ a lot between these products, so the lowest seat price isn't automatically the cheapest option for your usage.
| Product | $/user/mo | $/month | $/year | Include |
|---|
If this page is useful, the paid kit goes further. It has a unit economics model covering the same 24 models, cost allocation and tagging policy templates, monitoring queries for AWS, Azure and GCP billing data, a chargeback model, a maturity self-assessment, and a FOCUS™ converter that turns AI vendor usage exports into FOCUS-structured data at the revision you choose (1.0 to 1.4), so API spend can join the dataset you already build from the clouds' native FOCUS exports. The spreadsheet formulas are machine-checked, and each query notes which columns to verify against your own schema, since billing exports differ between accounts.
Instant download. VAT and sales tax handled at checkout.