エクサベース AI、OpenAI最新モデル「GPT-5.6 Sol」提供開始

Exa Enterprise AI, a subsidiary of ExaWizards (listed on the Tokyo Stock Exchange under ticker 4259), announced the launch of OpenAI's latest model GPT-5.6 Sol through exaBase AI, a corporate generative AI service that holds the market's top share for business AI applications. The company serves approximately 1,700 corporate users since the service's paid launch in June 2023.

GPT-5.6 Sol is OpenAI's flagship model in the GPT-5.6 series, engineered specifically for advanced professional tasks. The model introduces two key capabilities: "max reasoning effort," which allocates additional computational resources to deep reasoning on complex problems, and "ultra mode," which orchestrates multiple sub-agents in parallel to divide labor and accelerate solution of complicated tasks. These features are designed to extend the model's effectiveness for long-duration autonomous work and handling large codebases. The model also strengthens safeguards against misuse, with enhanced safety measures implemented during development, particularly for cybersecurity-related tasks.

Existing exaBase AI customers gain access to GPT-5.6 Sol without requiring additional sign-up or special conditions — the new model is immediately available through the same interface and workflow as current features. The service itself is built around enterprise-grade security and compliance, offering features such as administrator-controlled usage monitoring, custom forbidden-word lists, and the ability to upload proprietary company data for use in generating responses tailored to each organization's internal information. These design choices address long-standing corporate concerns about data security and regulatory compliance when deploying generative AI tools.

exaBase AI has established itself as the market-leading corporate generative AI service, and the addition of OpenAI's GPT-5.6 Sol represents a step up in capability for its existing user base. The model is positioned specifically for high-complexity tasks — multi-step agent work, scientific reasoning, and advanced coding — where precision and efficiency are both critical. The new reasoning and agent features (max reasoning effort and ultra mode) directly address the enterprise need to handle large-scale, long-running automated workflows without sacrificing accuracy.

The service emphasizes enterprise security and compliance from the ground up, allowing organizations to upload proprietary data and control usage through administrative features like forbidden-word registration. The frictionless rollout to existing users — no additional signup required — lowers barriers to adoption of the new model's advanced capabilities. This positions exaBase AI to serve larger, more complex projects within its customer base without requiring them to switch platforms or negotiate new terms.

No comments yet. Be the first to share your thoughts!