Sam Altman Coordinates GPT-5.6 Approval with Commerce and Treasury Officials
During a Thursday interview with CNBC, Altman stated that the GPT-5.6 Sol model is 54% more token-efficient on agentic coding tasks and is “as good or better” than competing models currently on the market. Altman emphasized the company’s focus on enterprise value, noting, “Every enterprise now is thinking about spend and the value they’re getting in exchange for AI, and this is what we really want to do.”
The GPT-5.6 family is designed to match specific job requirements. GPT-5.6 Sol offers the highest reasoning ceiling, optimized for complex reasoning over large codebases and demanding agentic work. GPT-5.6 Terra serves as the “balanced default” for everyday interactive and agentic coding, while GPT-5.6 Luna provides a lightweight, cost-efficient option for faster tasks. These models are now rolling out in GitHub Copilot, where they are billed at provider list pricing under Usage Based Billing.
OpenAI Integrates ChatGPT Work with Slack and Google Drive for Workflow Automation
In conjunction with the model release, OpenAI unveiled “ChatGPT Work.” This new AI agent combines the capabilities of ChatGPT and Codex, allowing non-technical users to automate workflows. According to OpenAI, the tool can gather context from apps, files, and workflows to create finished materials such as documents, spreadsheets, presentations, and web apps. A “unified plugins directory” allows ChatGPT Work to connect to tools including Slack, Gmail, Google Drive, calendars, and CRMs.
The rollout is occurring globally, with immediate access provided to Mac and Windows users of the ChatGPT desktop app. On mobile and the web, Pro, Enterprise, and Edu users have priority access, with Plus and Business users receiving availability gradually. In GitHub Copilot, administrators must enable the policy for the GPT-5.6 models, as the feature is off by default.
Meta and SpaceX Compete with OpenAI by Launching Muse Spark 1.1 and Grok 4.5
OpenAI is positioning the GPT-5.6 suite as a lower-cost alternative to competitors amid industry-wide concerns regarding AI lab costs. The company is competing directly with other major players, including Anthropic, which recently launched its Claude Cowork agent. The broader sector remains highly competitive; on the same day as the OpenAI announcement, Meta unveiled its “Muse Spark 1.1” model, which the company described as its “strongest model for agentic and coding work yet.” Additionally, SpaceX, which acquired Elon Musk’s xAI earlier this year, launched “Grok 4.5” on the preceding Wednesday.
OpenAI, founded as a nonprofit in 2015, is currently valued at $852 billion by private investors. While reports have suggested the company is in talks with the Trump administration regarding a potential 5% government stake, Altman dismissed these claims as having a lot of inaccuracies. Regarding the prospect of a public offering, Altman stated, “I don’t know,” noting that while OpenAI has confidentially filed prospectuses with regulators, no official timeline for an IPO has been disclosed.
Altman expressed a vision for global access to AI, stating, “If you want broad access, which we do, and you have powerful models, you really want to be able to be confident in your safety claims, because otherwise the world is going to get uncomfortable very fast.” He added that the company hopes for a global regulatory approach, emphasizing that “everybody will get access” and that the U.S. is not expected to disproportionately benefit from the technology.
Find more reporting in our 科学＆テクノロジー section.
