Skip to content

Model pricing configuration

Cate’s Logged Usage estimates costs using Anthropic’s published retail API key pricing. If you’re on a subscription plan (Claude Pro or Max), an enterprise agreement, or you want to add a model that isn’t in Cate’s built-in list, you can adjust pricing in one of two places:

  • Settings → General — a savings multiplier slider, for Pro/Max subscribers who want a quick adjustment
  • Settings → Advanced — full JSON configuration for adding new models or setting per-token rates

Subscription savings slider (Settings → General)

Section titled “Subscription savings slider (Settings → General)”

Under General → Cost → Subscription savings, a slider lets you tell Cate how much cheaper your Claude subscription is compared to pay-as-you-go API pricing. It runs from 1x (no adjustment) to 50x.

Cate takes the retail-priced estimate and divides it by your chosen factor. If you set it to 5x, displayed costs are 1/5th of what the same token usage would cost at API key rates.

For example: a session Cate estimates at $1.00 retail would display as $0.20 at a 5x savings factor.

This is the quickest adjustment for Pro or Max subscribers. For zero-cost display or per-model customization, use the Advanced JSON config below.

Custom model pricing (Settings → Advanced)

Section titled “Custom model pricing (Settings → Advanced)”
  1. Open Settings (gear icon, bottom left)
  2. Click Advanced
  3. Find the Anthropic model list field — it accepts a JSON array

An example configuration is shown directly in the settings UI.

Each entry in the array defines one model and its token prices:

[
{
"id": "claude-opus-4-7-20251101",
"name": "Claude Opus 4.7",
"input_price_per_million": 15.00,
"output_price_per_million": 75.00,
"cache_write_price_per_million": 18.75,
"cache_read_price_per_million": 1.50
}
]
FieldTypeDescription
idstringThe model ID as it appears in JSONL session files (e.g. claude-sonnet-4-6)
namestringDisplay name shown in the Logged Usage UI
input_price_per_millionnumberPrice in USD per 1M input tokens
output_price_per_millionnumberPrice in USD per 1M output tokens
cache_write_price_per_millionnumberPrice in USD per 1M cache write tokens
cache_read_price_per_millionnumberPrice in USD per 1M cache read tokens

All prices are in USD. Use 0 for any tier if you want to exclude it from cost estimates (e.g., if you’re on a flat-rate subscription and input tokens are effectively free).

The quickest option is the Subscription savings slider in Settings → General. Set it to whatever factor best reflects your plan — most Pro/Max users land somewhere between 5x and 20x relative to API key pricing.

If you want to zero out estimates entirely (e.g. on an unlimited plan), use the Advanced JSON config and set all prices to 0:

[
{
"id": "claude-sonnet-4-6",
"name": "Claude Sonnet 4.6 (subscription)",
"input_price_per_million": 0,
"output_price_per_million": 0,
"cache_write_price_per_million": 0,
"cache_read_price_per_million": 0
}
]

This makes Logged Usage more useful as a measure of session efficiency (token counts, cache hit rate) without misleading cost figures.

When Anthropic releases a new model before Cate’s built-in list is updated, add it manually with its published pricing from platform.claude.com/docs/en/about-claude/pricing:

[
{
"id": "claude-new-model-id",
"name": "Claude New Model",
"input_price_per_million": 3.00,
"output_price_per_million": 15.00,
"cache_write_price_per_million": 3.75,
"cache_read_price_per_million": 0.30
}
]

The id must exactly match the model string in your JSONL session files. You can verify the model ID by opening a session’s Usage panel and checking the request table.

Enter your contracted per-token rates to make Logged Usage estimates reflect your actual agreement:

[
{
"id": "claude-opus-4-7-20251101",
"name": "Claude Opus 4.7 (enterprise)",
"input_price_per_million": 10.00,
"output_price_per_million": 50.00,
"cache_write_price_per_million": 12.50,
"cache_read_price_per_million": 1.00
}
]