Microsoft has made OpenAI’s GPT-5.6 Sol the default AI model for its employees. At the same time the company has told staff to stop tokenmaxxing. An internal memo from executive vice-president Jay Parikh, reported by 404 Media, says employees need to watch how they use tokens with GitHub Copilot. Tokenmaxxing means burning the maximum number of AI tokens. Parikh wrote that the company is not optimising for that. Instead it wants staff focused on outcomes that help customers and the business. The switch to GPT-5.6 Sol is meant to get more value from the token spend because the model is cheaper than some alternatives. From July 2026 each Microsoft division has an AI token budget target. Employees can track their own spending. Many engineers currently use hundreds to a few thousand dollars in tokens each month. The company says the changes are not meant to slow its AI-first goals.
Microsoft has made OpenAI’s GPT-5.6 Sol the default AI model for its employees. At the same time the company has told staff to stop tokenmaxxing. An internal memo from executive vice-president Jay Parikh, reported by 404 Media, says employees need to watch how they use tokens with GitHub Copilot. Tokenmaxxing means burning the maximum number of AI tokens. Parikh wrote that the company is not optimising for that. Instead it wants staff focused on outcomes that help customers and the business. The switch to GPT-5.6 Sol is meant to get more value from the token spend because the model is cheaper than some alternatives. From July 2026 each Microsoft division has an AI token budget target. Employees can track their own spending. Many engineers currently use hundreds to a few thousand dollars in tokens each month. The company says the changes are not meant to slow its AI-first goals.