We're introducing thinking credits—a new compute system that removes the usage limits holding back Fabric's AI capabilities.

Previously, we maintained protective limits on AI features to keep costs predictable. While this ensured reasonable pricing, it also constrained what Fabric could become. Thinking credits change this dynamic entirely.

A flexible foundation for AI compute

Thinking credits power every AI operation in Fabric. When you upload a video, we analyze the first three minutes of footage, generate transcripts, create embeddings, and produce summaries—consuming roughly 0.1 credits per upload. Conversations with the AI assistant, document processing, and automated workflows all draw from your credit balance based on their computational complexity.

This metering approach lets us offer significantly more powerful capabilities. Instead of hard limits, you have flexible capacity that scales with your actual usage.

Unlocking new possibilities

With thinking credits managing compute costs, we can now support workloads that were previously impossible:

Process thousands of hours of video content with deep analysis across visual and audio elements. Run extensive automated workflows that handle hundreds of documents daily. Engage with the most advanced reasoning models for complex, extended problem-solving sessions.

These capabilities represent just the beginning. Thinking credits provide the infrastructure for more sophisticated media analysis, advanced automation systems, and access to cutting-edge AI models as they become available.

How the system works

Every Fabric plan includes monthly thinking credits, with allocations varying by tier. Credits refresh each month but don't accumulate—unused credits expire at the end of your billing cycle.

For users requiring additional capacity, manual or automatic top-ups are available. Unlike monthly allocations, purchased credits persist indefinitely until consumed, providing flexibility for variable workloads or intensive projects.

Designed for different usage patterns

Most users will find their monthly allocation sufficient for typical workflows—uploading videos occasionally, using the AI assistant for research, processing documents as needed. For these users, thinking credits operate transparently in the background.

Power users can scale their usage dramatically. Whether you're analyzing extensive media libraries, running sophisticated automation, or conducting deep research with advanced models, thinking credits provide the headroom to work without artificial constraints.

Building for the future

Thinking credits represent more than a billing mechanism—they're infrastructure for Fabric's evolution. As AI capabilities advance and new models emerge, this system ensures we can integrate them seamlessly while maintaining sustainable economics.

Rather than raising prices universally, thinking credits let us keep base plans accessible while enabling users who need advanced capabilities to access them. This approach aligns costs with value while removing barriers to innovation.

The result is a platform that can grow with your needs, whether you're just getting started or pushing the boundaries of what's possible with AI-powered workflows.