GPT-5.6: OpenAI's New Model Family Redraws the AI Landscape

A New Chapter in the AI Race

In yet another move within the rapidly accelerating wave of AI releases, OpenAI has unveiled its new model family under the name GPT-5.6 — a suite designed to meet diverse needs spanning enterprise environments, everyday tasks, scientific research, and cybersecurity.

Three Models for Three Purposes

The GPT-5.6 family consists of three variants that differ in capability and price, each targeting a specific user segment:

• Sol: The flagship and most powerful model in the family, offering the optimal choice for intensive tasks such as coding and cybersecurity.
• Terra: A balanced option sitting midway between performance and cost, suited for everyday enterprise use.
• Luna: The budget-friendly model aimed at users seeking adequate performance at a lower price point.

Pricing per million tokens varies as follows: Sol at $5 for input and $30 for output; Terra at $2.50 for input and $15 for output; and Luna, the most affordable, at $1 for input and $6 for output.

Greater Efficiency, Lower Cost

CEO Sam Altman confirmed that the new models represent a significant leap in operational efficiency compared to previous releases. In statements to CNBC, he noted that the Sol model is 54% more token-efficient on coding tasks, translating to a tangible reduction in costs while maintaining or improving output quality.

Cybersecurity: A Core Advantage

Cybersecurity occupies a prominent place among GPT-5.6's standout features, with the company describing it as "its most powerful model in this domain to date." The model supports a broad range of defensive activities, including:

• Threat analysis and threat mapping.
• Code review, vulnerability detection, and remediation.
• Blue Teaming — simulating hypothetical attacks to uncover weaknesses before real adversaries can exploit them.

The model's cybersecurity capabilities previously drew the attention of the U.S. government, which sought to delay its release amid concerns over potential misuse.

ChatGPT Work: The New Enterprise Companion

Alongside the model launch, OpenAI announced a new tool called ChatGPT Work, targeting enterprise teams and available across desktop, web, and mobile platforms. The tool enables common administrative tasks such as drafting documents and correspondence, preparing spreadsheets, and building presentations — positioning it as a direct competitor to a range of office productivity applications.

The Competition with Anthropic: A Battle of Benchmarks

These launches come against the backdrop of intensifying competition with Anthropic, which has built a strong foothold in the enterprise market. OpenAI cites the Artificial Analysis Coding Agent Index to demonstrate its models' superiority, claiming that Sol outperforms Anthropic's Fable 5 by 2.8 points on the coding benchmark, while consuming less than half the tokens, running at more than twice the speed, and costing a third of the price.

The comparison extends to the rest of the lineup, with the announcement noting that Terra matches the performance of Fable 5, while Luna surpasses the Opus 4.8 model.

Availability and Access

The GPT-5.6 models are now immediately available through ChatGPT, the Codex tool, and OpenAI's API, opening the door for developers and businesses to leverage them in building their own applications without delay.

✦ بقلم فريق دروب أيديا