OpenAI Releases GPT-5.6 Sol, Terra, and Luna for ChatGPT, ChatGPT Work, Codex, API

OpenAI released its GPT-5.6 model family for ChatGPT, ChatGPT Work, Codex, and the OpenAI API, with a rollout starting globally and set to reach full availability over the next 24 hours.

The lineup is built around three tiers—Sol as the flagship model, Terra as the lower-cost everyday work option, and Luna as the fastest and most affordable model—and OpenAI says it is moving GPT-5.6 from a limited preview into general availability.

9to5Google

OpenAI also introduced a new ultra setting that coordinates multiple agents across parallel workstreams, and the company says ultra uses four agents by default.

For developers, OpenAI said 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.

Pricing for the API 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.

OpenAI’s GPT-5.6 launch also came with ChatGPT Work, described by OpenAI as an agent in ChatGPT that helps users take on more ambitious tasks by gathering information across apps and workflows to create finished materials like sheets, slides, docs, and web apps.

Thurrott reported that ChatGPT Work is rolling out today for those on ChatGPT Pro, Enterprise, and Edu plans, and it is coming to Plus and Business plans in the next few days, while also being available in the ChatGPT desktop app even on the free plan.

Breakingthenews.net

In parallel, Business Insider said OpenAI is merging its popular Codex AI coding tool into its ChatGPT desktop app and revealed a Work setting for the ChatGPT app.

The GitHub Blog said GPT-5.6 is now rolling out in GitHub Copilot, with GPT-5.6 Sol available to Copilot Pro+, Max, Business, and Enterprise SKUs and GPT-5.6 Terra and GPT-5.6 Luna available to Pro, Pro+, Max, Business and Enterprise SKUs.

Separately, SiliconANGLE reported that OpenAI introduced GPT-Live, a voice model series with GPT-Live-1 as the default option in paid ChatGPT plans and GPT-Live-1-mini powering the free tier, rolling out to iOS, Android and web versions of ChatGPT.

OpenAI said GPT-5.6 Sol sets new or near-frontier results across evaluations, highlighting 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 OpenAI product release page emphasized that GPT-5.6 launches with “our most robust safeguards to date,” describing layers of protections trained into the model with real-time checks, monitoring, and access calibrated to trust and risk.

OpenAI also described ultra as its highest-capability setting coordinating multiple agents across parallel workstreams, and it said ultra trades higher token use for stronger results and faster time-to-result on demanding tasks.

In the API, OpenAI added more predictable prompt caching, including explicit cache breakpoints and a 30-minute minimum cache life, with cache writes billed at 1.25x the uncached input rate and cache reads retaining a 90% discount on cached input.

OpenAI further said GPT-5.6 is used internally by OpenAI researchers for debugging systems, optimizing training, running experiments, and interpreting results.