GPT-5.6 Luna Gets an 80% Price Cut: Output Falls to $1.20 per 1M Tokens
OpenAI cuts GPT-5.6 Terra by 20% and Luna by 80%. Luna now costs $0.20 input and $1.20 output per 1M tokens, while Sol pricing and subscription plans stay unchanged.
Published 51 days ago. Content may be outdated.
OpenAI has cut prices for GPT-5.6 Terra and GPT-5.6 Luna.
Starting July 30, Terra input pricing is down 20%, while Luna input and output pricing are down 80%. Sol stays unchanged. ChatGPT and Codex subscription prices also stay the same, but Terra and Luna consume fewer credits inside subscription products.
Old vs. New Pricing
Per 1 million tokens, the standard API prices are now:
| Model | Old Input / Output | New Input / Output | Change |
|---|---|---|---|
| GPT-5.6 Sol | $5 / $30 | $5 / $30 | No change |
| GPT-5.6 Terra | $2.50 / $15 | $2 / $12 | -20% |
| GPT-5.6 Luna | $1 / $6 | $0.20 / $1.20 | -80% |
Luna’s new price is the standout: $0.20 per million input tokens and $1.20 per million output tokens. For batch classification, code generation, document handling, and multi-step agents, that moves it from “interesting to test” into “cheap enough to put into production by default.”
What Does the Cut Mean?
OpenAI says the lower prices reflect better efficiency in model routing, inference systems, production software, and context handling, and that the savings are being passed through to customers.
For developers, the practical split is now cleaner:
- Sol: complex reasoning, high-stakes decisions, and harder tasks
- Terra: everyday knowledge work and medium-complexity agents
- Luna: high-throughput, low-cost, repetitive work
One useful workflow is to let Sol handle planning and uncertainty first, then hand execution to Luna once the task is well defined. That is also the direction OpenAI is pushing for GPT-5.6 usage overall: different stages, different models.
One More API Change: Priority Processing Becomes Fast Mode
OpenAI also renamed API Priority Processing to Fast mode.
For GPT-5.6 Sol, Fast mode can run up to 2.5x faster than Standard processing, costs 2x the standard price, and does not change model intelligence. Existing requests that use service_tier: "priority" remain compatible and will map to Fast mode.
Will Individual Users Pay Less?
It depends on which product you use:
- API users: Terra and Luna bill at the new lower rates right away
- ChatGPT Work and Codex users: subscription prices stay unchanged, but Terra and Luna consume fewer credits
- Sol users: no pricing change in either API or subscription products
So this is not a direct ChatGPT Plus or Pro discount. It is a model pricing update that mainly affects API developers and teams with heavy usage.
Finally
The biggest change here is not that Sol got cheaper. It is that Luna has pushed high-quality model usage much lower on the cost curve. Terra is now the balanced everyday option, Luna is the scale option, and Sol remains the hard-problem model.
The real shift is that model selection is becoming workflow design. If a cheaper model can already finish more of the pipeline, you should probably re-test your routing strategy now rather than later.
Source: OpenAI: Advancing the price-performance frontier with GPT-5.6
More Articles
