Microsoft ships Work IQ API GA with consumption-based Copilot Credits billing

Microsoft brought its Work IQ API to general availability on June 16, introducing consumption-based billing through 'Copilot Credits' unified across Copilot Studio and AI services. The shift to metered, consumption-based pricing signals Microsoft's intent to monetize agentic AI usage directly—charging for what agents actually consume rather than bundling it into flat subscriptions. In parallel, Microsoft 365 E5 became a prerequisite for new Agent 365 purchases, tightening the licensing requirements around its agent platform.

The billing changes have proven contentious. The same Copilot Credits model that powers Work IQ has drawn sharp criticism over Copilot Cowork's token consumption, with a single agentic coding session reportedly burning $30–$40 in credits—three to four times a Pro subscriber's monthly allotment. Reddit threads accused Microsoft of 'building features that encourage high token consumption, then penalizing users,' a backlash that helps explain Microsoft's reported interest in cheaper DeepSeek-based models.

The E5 requirement for Agent 365 is a classic Microsoft bundling play, pushing customers toward its highest enterprise tier to access agentic capabilities. Combined with consumption billing, it reflects a deliberate strategy to capture more enterprise spend as agents become central to productivity workflows. Azure Horizon DB also entered preview, supporting up to 3,072 vCores—aimed at the data-intensive backends these agents require.

Competitively, Microsoft's metered-credits approach contrasts with rivals still experimenting with pricing, and it bets that enterprises will accept usage-based costs for sufficiently valuable agents. The risk is exactly the backlash already visible: if costs feel unpredictable or punitive, customers balk. Watch how Work IQ adoption proceeds, whether Microsoft adjusts credit allotments in response to complaints, and how the multi-model strategy (potentially including DeepSeek) lowers the per-task cost that's driving the controversy.