Token mix by model family

The token mix groups monthly AI token telemetry by model family. Claude Opus families are labeled by their model family name, so Opus 4.8, Opus 4.7, Opus 4.6, GPT-5.5, and similar families each appear together.

The 100% stacked bar chart shows monthly shares over time. It keeps the ten families with the highest average monthly share across the reporting period visible and folds the remaining families into Other. Each month's stack sums to 100% of token volume.

This metric measures token volume, not cost. Model prices vary, so token share and spend share should be interpreted separately.

Sample and interpretation

The public aggregate snapshot includes model-family shares, but does not include employee counts, organization counts, or total token counts.

These are aggregate observations from Weave customer telemetry, not a controlled experiment or a representative sample of every engineering team. Changes in the observed cohort and connected telemetry can affect the series. The data alone does not establish that AI caused a change in output.

Before comparing metrics or reporting periods, check the benchmark populations, units, and exclusions.

For the wider measurement framework, read how to calculate AI coding ROI, including cost attribution, quality, and baseline comparisons.