Open models take the top spend share on shell execution and tool dispatch

OpenRouter reported that DeepSeek V4 Pro led spending for shell execution and Kimi K3 led spending for tool dispatch in its August 6 snapshot. These categories concern running commands and invoking tools. The attached chart shows DeepSeek V4 Pro at 18.0% of shell-execution spend, just ahead of GPT-5.4 at 16.1%, so the lead is narrow and far from a majority. The result shows where customers allocated money for those workloads on this platform. It does not measure how often the models completed tasks successfully, so a spending lead should not be read as an accuracy ranking.