What is tokenomics?
Tokenomics is the economics of AI tokens: what each token costs, how it trades off against performance, and the energy, carbon and water behind it.
Why it matters
AI providers charge by the token, so token volumes drive AI spend. Research on LLM services uses tokenomics to describe the balance between performance and cost. Model choice, prompt and answer length, caching and batching all change what each token costs.
Most changes that cut token costs also cut energy. That makes tokenomics a shared concern for FinOps and sustainability teams, who can work from the same usage data.
An example
Take a team sending simple classification jobs to a frontier model in real time. Moving them to a smaller model lowers the price per token and the energy per token. Shorter outputs, cached system prompts and batch APIs for non-urgent work each lower spend and electricity together.
How Greenpixie uses it
Greenpixie extends tokenomics to sustainability by putting energy, carbon and water per token next to AI spend in GPX AI. Teams see both when they choose a model or a serving option. GPX AI breaks this down by model and by token type, so a team can see which models and which kinds of token drive both the bill and the footprint.
Price also works as a check. Greenpixie's AI methodology tests whether the implied electricity cost of a token stays within plausible bounds against what the provider charges for it. If it takes too large a share, the estimate is flagged for review. The check is weaker where a provider prices close to its cost, so it is used alongside external comparisons.