OpenAI, the developer of ChatGPT, said on the 9th (local time) that it officially launched its latest artificial intelligence (AI) model, "GPT-5.6."
OpenAI, at the request of the U.S. government, had pre-released the GPT-5.6 lineup to only some agencies over the past two weeks and opened it to the public on this day. GPT-5.6 consists of the high-performance top-tier model Sol, the second-tier model Terra, and the cost-efficient Luna.
OpenAI said the new model not only set top performance records in areas such as coding, knowledge work, and scientific research, but can perform the same tasks with fewer tokens and lower expense than rival top-tier models, improving "performance per dollar."
Sam Altman, OpenAI chief executive officer (CEO), said in a CNBC interview on this day, "GPT-5.6 Sol improved token efficiency for agentic coding tasks by 54%." A token is the basic unit an AI model uses to process information and generate answers. Altman noted, "corporations are focused on what value they are getting in return for their AI expenditure," emphasizing that GPT-5.6 delivers superior performance-to-expense efficiency.
OpenAI called the top-tier model Sol "the best coding model ever." It also said its cybersecurity capability rivals Anthropic's "Claude Mythos 5."
On Terminal-Bench 2.1, which measures coding ability in a terminal environment, Sol scored 88.8%, narrowly topping Mythos 5's 88%. On "CyberGym," which evaluates cybersecurity capability, it scored 84.5%, surpassing Mythos 5's 83.8%. On the "Agent's Last Exam" (ALE) metric, which measures practical agent capability, it posted 52.7%, far ahead of Fable 5 (40.5%) and Opus 4.8 (45.2%).
However, on SWE-Bench Pro, which measures programming language coding ability, it scored 64.6%, falling short of Mythos 5 (80.3%) and Opus 4.8 (69.2%).
Artificial Analysis, an AI model evaluation firm, rated GPT-5.6 Sol at 59 points, placing it second behind Fable 5 (60 points). GPT-5.6 Terra received 55 points, ranking fourth after Opus 4.8 (56 points).
Accordingly, "Grok 4.5" from SpaceXAI, which debuted in fourth place the previous day, fell two spots to sixth place in just one day.
API (application programming interface) prices for GPT-5.6, based on input and output per 1 million tokens, are $5 and $30 for Sol, $2.5 and $15 for Terra, and $1 and $6 for Luna.