Sidekick Orchestration
// The Sidekick Skim · July 25 - 31, 2026

Cheaper AI Does Not Mean Cheaper Work

Luna's API price fell 80%, but list price is only one part of operating cost. Measure accepted outputs, review time, retries, integration, and operating complexity before you move a workflow.

In this edition
Read this if you read nothing else
This week's decision

Reprice one repeatable workflow on ten representative jobs and count only outputs that meet the quality bar.

What changed

Luna fell to $0.20 per million input tokens and $1.20 per million output tokens, one-fifth of its prior list price.

What it means

Routine, high-volume work became cheaper at list price, but the operating question is still cost per accepted result after review and retries.

Safe to ignore

Do not migrate from list price or provider benchmarks alone. A cheaper run can still cost more when quality, review, retries, integration, or operating complexity gets worse.

Only where it applies
  1. If you run a repeatable workflow through an API30 minutes

    Reprice one real task

    Run the same ten representative jobs on your current model and Luna. Count only outputs that meet the quality bar, then include review and retry time.

    Limit: Do not move customer-facing, regulated, or high-cost-of-error work on list price alone.
01

The cost curve moved again

Relevant if API usage or agent credits are a material operating cost.

PricingCompany pricing; current API documentationOpenAI cut Luna by 80% and Terra by 20%Luna moved from $1 input and $6 output per million tokens to $0.20 and $1.20. Terra moved from $2.50 and $15 to $2 and $12.Applies to: Teams running high-volume, repeatable API or Codex workflows

For a workflow using 10 million input tokens and 2 million output tokens, Luna's list-price cost falls from $22 to $4.40, a saving of $17.60. OpenAI kept ChatGPT and Codex subscription prices and quota budgets unchanged, but says Luna and Terra now consume fewer credits.

Your decision

Reprice one high-volume workflow, then divide the full cost by outputs that met your quality bar. A price falling to one-fifth is worth testing, not an automatic migration. Long prompts above 272,000 input tokens carry higher rates.

Sources: OpenAI, OpenAI API docs

Evidence note: Sources checked August 14, 2026. OpenAI's performance and cost comparisons are company claims. Published API prices are list prices and do not include review time, retries, integration, or operating complexity.

// Get the next edition

Get the next Skim.

New editions by email when they clear the evidence and usefulness checks.

// Need help applying this decision?

Bring one decision. Leave with a next step.

Use the 30-minute call to work through that specific model, workflow, or risk choice.