--- title: "OpenAI cuts GPT-5.6 prices: Luna 80% lower, Terra 20% lower" description: "OpenAI has cut GPT-5.6 Luna prices by 80% and Terra by 20%, and introduced Fast mode for Sol. Review API prices, availability, and practical model selection." url: "https://kylon.io/blog/gpt-5-6-pricing-update-july-2026" language: "en" --- [HOME](https://kylon.io/)\>[Blog](https://kylon.io/blog)\>OpenAI cuts GPT-5.6 prices: Luna drops 80%, Terra drops 20% AI News5 min read # OpenAI cuts GPT-5.6 prices: Luna drops 80%, Terra drops 20% OpenAI has reduced GPT-5.6 Luna pricing by 80% and Terra pricing by 20%, while adding Fast mode for Sol in the API. Here is what changes for teams building production workflows. ![Clara](https://kylon.io/v2/clara-avatar.png) ClaraGTM & Content July 30, 2026 [A narrated line-art explainer of OpenAI cuts GPT-5.6 prices: Luna drops 80%, Terra drops 20%, with subtitles](https://static.purework.app/files/7e10e1af23724909096a610ffba277cd/explainer-gpt-5-6-pricing-update-july-2026.mp4) ## The short version On July 30, OpenAI reduced the price of **GPT-5.6 Luna by 80%** and **GPT-5.6 Terra by 20%**. It also introduced **Fast mode** for GPT-5.6 Sol in the API, replacing Priority Processing. This is not a new-model release. It is a material change to the cost and speed profile of the GPT-5.6 family, particularly for teams running high-volume production workflows. ## What changed | Model or feature | Change | Current API price | | --- | --- | --- | | GPT-5.6 Luna | 80% lower price | $0.20 input / $1.20 output per 1M tokens | | GPT-5.6 Terra | 20% lower price | $2 input / $12 output per 1M tokens | | GPT-5.6 Sol Fast mode | Up to 2.5x Standard speed | 2x Standard price | OpenAI positions Luna as the fast, lower-cost option for high-volume work. Terra is the balanced model for everyday tasks. Sol remains the higher-capability option for demanding work. ## Fast mode replaces Priority Processing Fast mode is available for GPT-5.6 Sol in the API. OpenAI says it can deliver up to 2.5x the speed of Standard processing, at twice the price, with no change in intelligence. Existing API requests tagged `priority` will continue to work and will use Fast mode. That makes the trade-off explicit: teams can pay for lower latency when the workflow justifies it, rather than applying premium processing to every request. ## Availability Terra and Luna remain available in the OpenAI API, ChatGPT Work, and Codex. In ChatGPT Work and Codex, Free and Go users can access Terra. Plus, Pro, Business, and Enterprise users can choose both Terra and Luna. OpenAI says subscription prices and quota budgets are unchanged, while Terra and Luna consume fewer credits. ## Why this matters for production workflows The practical question is not which model has the highest capability in isolation. It is which model is appropriate for each stage of work. A team could use Sol to resolve uncertainty and set a plan, then route well-specified implementation, testing, document classification, or extraction work to Luna. Terra can sit between those extremes for routine knowledge work. The new pricing widens the gap between those cost profiles. That approach is easier to operate when the workflow, source context, and approval rules live in one place. Kylon lets teams bring AI agents into the same workspace as their rooms, files, rules, and recurring workflows, so model selection can be connected to the work itself rather than reinvented in each prompt. ## Bottom line - Luna is now 80% less expensive. - Terra is now 20% less expensive. - Sol gets a faster API processing option through Fast mode. - Teams should review high-volume jobs and separate routine execution from higher-stakes reasoning. Read OpenAI's official announcement for the complete pricing and availability details. Your first company harness. Where humans and agents run your business together. [Get started](https://app.kylon.io/sign-up) ## Keep reading [All posts](https://kylon.io/blog) [ ### Turn a Mac mini into a teammate that works 24/7 Claude Code and Codex on an always-on Mac mini, working with your team in Kylon while your laptop is closed. The setup we repeat for customers and for ourselves. Kylon TeamNovember 1, 2026 ](https://kylon.io/blog/mac-mini-24-7-agents)[ ### The AI-native tax The hidden cost of putting AI to work, and who ends up paying it. Quinn LengOctober 29, 2026 ](https://kylon.io/blog/the-ai-native-tax)[ ### Agent governance for teams: start with the boundaries a prompt can't move A customer running department and workflow agents asked what happens if someone jailbreaks one. The answer depends on where the rules live. The four boundaries Kylon enforces whatever the agent is told, how agent memory works, and how we would set up agents around them. Kylon TeamOctober 27, 2026 ](https://kylon.io/blog/agent-governance-hard-boundaries)