OpenAI has released GPT-5.6, its new frontier model family for ChatGPT, ChatGPT Work, Codex, and the OpenAI API. The rollout starts globally today and is set to reach full availability over the next 24 hours. The family includes three tiers: Sol as the flagship model, Terra as the lower-cost everyday work option, and Luna as the fastest and most affordable model.
The launch moves GPT-5.6 from a limited preview into general availability. OpenAI is positioning the series around higher intelligence per token, lower estimated cost for complex work, and stronger agentic performance across coding, knowledge work, cybersecurity, science, design, and internal research workflows. Sol is the top-tier model, while Terra and Luna are intended to make the same generation available at lower cost and latency.
A major addition is the new ultra setting, which coordinates multiple agents across parallel workstreams for demanding tasks. OpenAI says ultra uses four agents by default, trading higher token use for stronger results and faster completion on complex work. Developers can build similar workflows through the multi-agent beta in the Responses API, while Programmatic Tool Calling lets GPT-5.6 write and run in-memory JavaScript to coordinate tools, call them in parallel, use loops and conditions, and process intermediate results before returning an answer.
GPT-5.6 is also aimed at professional artifact generation. OpenAI says the model can create editable presentations, documents, spreadsheets, interfaces, visual explanations, and frontend prototypes with stronger layout judgment and closer adherence to reference files. In ChatGPT Work, the model is designed to handle source material from documents and connected work apps, then convert it into shareable outputs. This puts GPT-5.6 directly into OpenAI’s broader push to make ChatGPT a work-execution environment rather than just a conversational assistant.
For developers, GPT-5.6 arrives with Sol, Terra, and Luna in the API. Pricing is set at $5 input and $30 output per 1 million tokens for Sol, $2.50 input and $15 output per 1 million tokens for Terra, and $1 input and $6 output per 1 million tokens for Luna. The release also adds more predictable prompt caching, explicit cache breakpoints, and a 30-minute minimum cache life. Cache writes are billed at 1.25x the uncached input rate, while cache reads retain a 90% discount on the cached input rate. Access depends on product and plan. In ChatGPT, Plus, Pro, Business, and Enterprise users get GPT-5.6 Sol through medium- and higher-effort settings, while Pro and Enterprise users can select Sol Pro for complex tasks. In ChatGPT Work and Codex, Free and Go users get Terra, while Plus, Pro, Business, and Enterprise users can choose Sol, Terra, and Luna with effort controls. The max setting is available to all users with access to GPT-5.6 in ChatGPT Work and Codex. Ultra is available to Pro and Enterprise users in ChatGPT Work and to Plus and higher plans in Codex.
OpenAI says GPT-5.6 Sol sets new or near-frontier results across several evaluations. The company highlights gains in coding with Terminal-Bench 2.1 and DeepSWE; knowledge work with BrowseComp and OSWorld 2.0; cybersecurity with ExploitBench, ExploitGym, and SEC-Bench Pro; and scientific workflows with GeneBench Pro, LifeSciBench, and chemistry-related evaluations. The company also says GPT-5.6 is now used internally by OpenAI researchers for debugging systems, optimizing training, running experiments, and interpreting results.