Last month, OpenAI announced a limited preview of its GPT-5.6 model series, giving access only to a small group of trusted partners. Today, OpenAI is expanding access to all three models in the family: GPT-5.6 Sol, GPT-5.6 Terra, and GPT-5.6 Luna.
GPT-5.6 Sol is the flagship model, built for complex reasoning, coding, scientific work, cybersecurity, and long-running agentic tasks. GPT-5.6 Terra is the balanced model in the lineup; OpenAI says it delivers performance competitive with GPT-5.5 while being twice as cheap. GPT-5.6 Luna is the smallest and most cost-efficient model in the series, aimed at faster, lower-cost workloads.
With GPT‑5.6, OpenAI has introduced a new "max reasoning effort" setting, which can be selected for the Sol model to further improve reasoning performance. Additionally, a new "ultra mode" goes beyond the capabilities of a single agent by utilizing subagents to accelerate complex reasoning workflows.
GPT-5.6 family benchmark highlights:
- Agents’ Last Exam: GPT-5.6 Sol scored 53.6, which is 13.1 points higher than Claude Fable 5 using adaptive reasoning. Even at medium reasoning, Sol beat Fable 5 by 11.4 points at about one-quarter of the estimated cost.
- Artificial Analysis Intelligence Index: GPT-5.6 Sol with max reasoning came within one point of Fable 5, while completing tasks 61% faster at roughly half the estimated cost.
- Artificial Analysis Coding Agent Index: GPT-5.6 Sol with max reasoning scored 80, setting a new state of the art. This is 2.8 points above Fable 5, while using less than half the output tokens, less than half the time, and about one-third lower estimated cost.
- GPT-5.6 Sol set new state-of-the-art results on Terminal-Bench 2.1 and DeepSWE v1.1, both focused on complex engineering and real-codebase workflows.
- In BrowseComp, GPT-5.6 Sol achieved 92.2%, which is a new state-of-the-art result for agentic browsing tasks.
- In OSWorld 2.0, GPT-5.6 Sol scored 62.6%, surpassing Opus 4.8 while using 85% fewer output tokens.
- In ExploitBench 1, GPT-5.6 Sol scored 73.5%, compared with 47.9% for GPT-5.5 at a comparable output-token budget.
- In ExploitGym 2, GPT-5.6 Sol nearly doubled GPT-5.5’s peak pass rate, rising from 15.1% to 24.9% under the two-hour cap. With six hours, it reached 33.7%.
- In SEC-Bench Pro, GPT-5.6 Sol scored 71.2%, compared with 45.8% for GPT-5.5, with improved latency.
Pricing for the GPT-5.6 family is as follows:
- GPT-5.6 Sol: $5 per million input tokens and $30 per million output tokens
- GPT-5.6 Terra: $2.50 per million input tokens and $15 per million output tokens
- GPT-5.6 Luna: $1 per million input tokens and $6 per million output tokens
- OpenAI will also be offering GPT-5.6 Sol in a fast mode through Cerebras, delivering speeds of up to 750 tokens per second.
GPT‑5.6 will be available over the next 24 hours across ChatGPT, Codex, and the OpenAI API. ChatGPT Plus, Pro, Business, and Enterprise users can access GPT‑5.6 Sol through medium and higher effort settings. ChatGPT Pro and Enterprise users can select GPT‑5.6 Sol Pro for complex tasks.
ChatGPT Free and Go users can access GPT‑5.6 Terra. ChatGPT Plus, Pro, Business, and Enterprise users can choose between GPT‑5.6 Sol, Terra, and Luna and set a reasoning effort level for each. The new max mode is available to all users with access to GPT‑5.6 in ChatGPT Work and Codex. In ChatGPT Work, GPT-5.6 Sol ultra is available to Pro and Enterprise users. And in Codex, the new ultra mode is available to Plus and higher plans.