OpenAI released the GPT-5.6 model family on July 9, 2026, in three tiers: GPT-5.6 Sol, GPT-5.6 Terra and GPT-5.6 Luna. According to OpenAI’s announcement, Sol is “our flagship” and Terra is “a balanced model for everyday work,” while the API changelog positions Luna “for efficient, high-volume workloads.” All three are available in the API, and the tiers are spread across ChatGPT, ChatGPT Work and Codex by plan.
On this page
Three tiers
OpenAI explains the naming this way: “The number identifies the generation, while Sol, Terra, and Luna are durable capability tiers that can advance on their own cadence.” The changelog describes Sol as built “for frontier capability” and Terra “for a balance of intelligence and cost.”
The API model IDs are gpt-5.6-sol, gpt-5.6-terra and gpt-5.6-luna.
Who gets which model
- ChatGPT: “Plus, Pro, Business, and Enterprise users access GPT‑5.6 Sol through medium and higher effort settings.”
- ChatGPT Work and Codex: “Free and Go users access GPT‑5.6 Terra. Plus, Pro, Business, and Enterprise users can choose among GPT‑5.6 Sol, Terra, and Luna.”
- API: “Developers can access Sol, Terra, and Luna through the OpenAI API.”
New API features
The changelog lists several additions that shipped with GPT-5.6:
- Programmatic Tool Calling
- explicit prompt caching controls
- persisted reasoning, a
maxreasoning effort level and Pro mode - multi-agent orchestration, in beta in the Responses API
Benchmark results for GPT-5.6 Sol
All scores below were published by OpenAI on the GPT-5.6 page and are its baseline results. OpenAI separately reports “Ultra” results that use parallel agents.
| Benchmark | GPT-5.6 Sol |
|---|---|
| Terminal-Bench 2.1 | 88.8% |
| BrowseComp | 90.4% |
| OSWorld 2.0 | 62.6% |
| Agents’ Last Exam | 52.7% |
| SEC-Bench Pro | 71.2% |
With parallel agents, OpenAI reports 91.9% on Terminal-Bench 2.1 and 92.2% on BrowseComp. Lab-published results for GPT-5.6 and later OpenAI models are tracked on our benchmark leaderboard.
What it changes
The Sol, Terra and Luna tier names carried into the GPT-6 family in September, when OpenAI released GPT-6 Sol and GPT-6 Luna, which fits OpenAI’s description of the tiers as “durable.” For developers, GPT-5.6 meant three model IDs in one generation, plus the max reasoning effort level, Programmatic Tool Calling and explicit prompt caching controls in the API. For Free and Go users, GPT-5.6 Terra is the model available in ChatGPT Work and Codex.
All three models are listed on our AI model release timeline, and API service health is tracked on the OpenAI API status page.
