OpenAI released its GPT-5.6 model family on July 9, 2026, giving ChatGPT, Codex, ChatGPT Work and API users a new three-tier lineup: Sol, Terra and Luna.
The simplest way to understand the launch is that OpenAI is separating capability levels more clearly. Sol is the flagship model for harder work, Terra is the balanced option for everyday tasks, and Luna is the fastest and lowest-cost tier.
This is not only a model-name update. OpenAI says GPT-5.6 is available across ChatGPT, Codex and the OpenAI API, with rollout timing and model access depending on product and plan. The company is also tying the release to more capable agent workflows, including ChatGPT Work and Codex features that can handle longer projects.
The short answer
Use Sol when the task is hard, risky or expensive to redo. Use Terra when you want a capable default for normal professional work. Use Luna when speed and cost matter more than maximum reasoning depth.
OpenAI's own release describes GPT-5.6 as a family designed for stronger performance per dollar. The company says Sol improves results across coding, knowledge work, cybersecurity and science, while Terra and Luna are meant to make newer capability available at lower cost.
What changed
The new model family has three durable tiers:
- Sol: OpenAI's flagship GPT-5.6 model, aimed at complex coding, research, science, cybersecurity, computer use and design work.
- Terra: A lower-cost balanced model for everyday professional work, including ChatGPT Work and Codex use cases.
- Luna: The fastest and most affordable GPT-5.6 model for high-volume tasks where cost and latency matter.
OpenAI also introduced higher-effort settings around the release. In ChatGPT Work and Codex, eligible users can use a `max` effort setting. An `ultra` mode is available in ChatGPT Work for Pro and Enterprise users and in Codex for Plus and higher plans, according to OpenAI's availability notes.
Who gets access
For ChatGPT, OpenAI says GPT-5.6 Sol is rolling out to eligible paid plans. The Help Center says Free, Go and logged-out users are not included for Sol in ChatGPT, and that managed workspace access can depend on administrator settings.
For ChatGPT Work and Codex, OpenAI says Free and Go users can access GPT-5.6 Terra, while Plus, Pro, Business and Enterprise users can choose among Sol, Terra and Luna. Developers can access all three through the OpenAI API.
What it costs in the API
OpenAI lists GPT-5.6 API prices per 1 million tokens as $5 input and $30 output for Sol, $2.50 input and $15 output for Terra, and $1 input and $6 output for Luna. The release also adds more predictable prompt caching, including explicit cache breakpoints and a 30-minute minimum cache life.
That pricing makes model routing more important. A developer might use Luna for draft transformations or data cleanup, Terra for normal coding and research assistance, and Sol for final review, difficult debugging, security analysis or work where a mistake is costly.
What to watch
The biggest question is how the new tiers perform in real use once more people have access. Benchmarks are useful, but users will care about everyday reliability: whether the model follows instructions, handles files and tools cleanly, and finishes long tasks without requiring constant correction.
For most readers, the practical takeaway is simple: do not treat the three model names as status labels. Treat them as routing choices. Pick the smallest model that can do the job well, then move up when the task has higher stakes or needs deeper reasoning.
