Sample data. Not a customer bill.

Model comparison

Planned

Cost, latency and measured quality on this organisation's traffic.

Figures follow the shell: Last 30 days · All projects · All environments.

Sample catalogue prices, not a live price book

Cost and requests are this organisation's traffic. Latency is the request sample in the window. Quality scores are catalogue-wide, not measured on this organisation's traffic. Price history is the versioned host price, which is never overwritten.

Models in this view

Spend

Price versions

Kept side by side

Next step

Shadow test

A comparison is not a switch

Shadow-test this

Spend and requests are this organisation's traffic. Quality scores are catalogue-wide, not measured on this traffic.

GPT-4.1	OpenAI	$11,995.56	1,226	683 ms	general 0.90	$2.00 / MTok	$8.00 / MTok	2026-09-01

1 rows · page 1 of 1

Price trend

Output price per million tokens from each host's price versions. Falling prices are why every audit keeps the version it used.

Parezen Index

View as table

version	openai_gpt_4_1_openai
2026-06-01	10
2026-09-01	8

Versioned host prices

GPT-4.1	OpenAI	2026-06-01	$2.50	$0.25	$10.00
GPT-4.1	OpenAI	2026-09-01	$2.00	$0.20	$8.00

2 rows · page 1 of 1

OpenAIGPT-4.1