OpenAI Unveils GPT-5.6 Sol Series: Flagship AI with Advanced Reasoning and Ultra Mode for Complex Workflows
OpenAI has launched a limited preview of its next-generation GPT-5.6 series, featuring the flagship Sol model alongside the Terra and Luna variants. This series introduces “Ultra Mode” and enhanced agentic capabilities designed to revolutionize professional workflows in coding, biology, and cybersecurity.
RMN Digital AI Desk
New Delhi | June 27, 2026
OpenAI Sets New Benchmark with GPT-5.6 Sol Series
On June 26, 2026, OpenAI announced the limited preview of its most powerful AI series to date: GPT-5.6. The lineup is led by Sol, the flagship model designed for high-intensity reasoning; Terra, a balanced model for daily professional use that performs similarly to GPT-5.5 at half the cost; and Luna, a fast, low-cost option for broad accessibility.
Advancing Reasoning with “Ultra Mode”
The GPT-5.6 series introduces a significant architectural shift with a new “max reasoning effort” setting, allowing Sol to dedicate more time to deep, complex problems. Furthermore, OpenAI is debuting “Ultra Mode,” which enables the AI to move beyond a single agent by utilizing subagents to accelerate and coordinate intricate tasks.
GPT-5.6 Sol isn’t just a model; it’s a multi-agent system designed to conquer the most complex coding and security challenges.
These improvements are reflected in recent benchmarks. GPT-5.6 Sol Ultra achieved a record-breaking 91.9% on Terminal-Bench 2.1, demonstrating superior planning and tool coordination in command-line environments. In biological research, Sol outperformed GPT-5.5 on GeneBench v1 while maintaining higher token efficiency.
A Frontier for Cybersecurity
Sol represents a “performance-efficiency frontier” shift for security professionals. It is specifically optimized for defensive tasks such as code review, patch development, and vulnerability research. During testing on ExploitBench, Sol proved competitive with rival models while requiring only about one-third of the output tokens. While Sol is adept at identifying bugs in browsers like Chromium and Firefox, OpenAI noted it did not autonomously produce functional full-chain exploits during evaluations, emphasizing its role as a tool for defenders rather than attackers.
With the introduction of Terra and Luna, OpenAI is delivering flagship-level intelligence at a fraction of the previous cost and latency.
Safety and Phased Availability
In a notable shift in release strategy, OpenAI has coordinated with the U.S. government to implement a phased rollout. At the government’s request, the preview is currently limited to a small group of trusted partners. This follows the development of the most robust safety stack in the company’s history, featuring layered safeguards such as real-time misuse classifiers and account-level monitoring.
“We don’t believe this kind of government access process should become the long-term default,” OpenAI stated, though they acknowledged it as the strongest path toward broader availability while finalizing the Cyber Executive Order framework.
The company plans to make the Sol, Terra, and Luna models generally available in the coming weeks.
