GitHub quietly updated the rate limit multipliers for Claude models inside Copilot, and the numbers are steep: Claude Sonnet now draws 9 premium requests per query, and Claude Opus draws 27. Most teams using these models had no idea until they started hitting their monthly caps faster than expected.
Copilot's billing works on a "premium requests" pool - each plan includes a monthly allowance, and different AI models pull from it at different rates. A standard query on cheaper models might count as 1 request. The new multipliers mean that switching to Claude doesn't just use more of that pool, it drains it at a rate that makes the math fall apart quickly. A team with 300 premium requests per month now gets roughly 33 Sonnet queries or 11 Opus queries before hitting their limit. That's not a usable workflow for any team doing sustained work.
The change appeared in GitHub's rate limits documentation without a corresponding announcement, email, or in-product warning. Developers discovered it by reading the fine print, not from any official communication. GitHub updated a multiplier table, and teams were left to figure out the downstream impact on their own.
The Practical Cost for Teams
For organizations that specifically chose Copilot to access Claude's capabilities - Opus in particular is among the strongest available models for complex coding tasks - this is a material pricing change disguised as a documentation update. The monthly cost-per-useful-query just jumped significantly.
This also affects how Claude Code users might think about routing work. Accessing Claude through Copilot's enterprise plans carries this multiplier overhead on top of the subscription cost. Direct API access through Anthropic prices by token (roughly per word processed), which is predictable and doesn't have hidden multipliers that GitHub can change without notice.
GitHub hasn't explained the reasoning publicly. It's possible the change reflects the actual inference cost difference between models, but the absence of any announcement makes it hard to evaluate. What's clear is that the effective price of using Claude through Copilot has changed substantially, and the teams most affected are the ones who relied on the original rate structure when deciding to use the product.