How Large Is the Open Model Economy Today?

The Interconnects Dashboard aggregates data on open model adoption and performance from across the web. Although the dashboard is freely available, measuring open model adoption remains difficult. Available indicators include noisy platform metrics, startup marketing figures, and imperfect public evaluations. Understanding the overall direction therefore requires interpreting several incomplete sources together.

These research notes provide current estimates of the open model economy, with a focus on the market for open model inference. The central questions are whether that market is growing and how large it is relative to services from OpenAI and Anthropic.

Measuring Open Model Inference

The main indicators come from the open model inference startups Baseten, Fireworks, and Together AI. Another useful source is OpenRouter, which reports the share of its token volume associated with open models through its API.

Hyperscalers, including Amazon Bedrock and Azure’s model garden, may also account for substantial usage. However, their public reports do not appear to break usage down by model category. The performance of the specialist startups is more directly connected to the development of open model inference as a distinct market.

Public Token-Volume Disclosures

Public disclosures from open model startups show near-exponential growth in token volume over time. The figures are presented on a logarithmic scale.

This growth predates the increase expected after GLM 5.2’s step change in open agents. At its first company conference, Fireworks announced that it was serving more than 40T tokens per day to customers’ customized models. That figure represents only one part of the company’s inference business, a segment Fireworks said had grown by more than two times since the summer.

Based on that disclosure, the company’s total volume in the fall could exceed 100T tokens per day when it provides its next update.

Comparing Open and Closed Model Volume

One way to interpret these figures is to compare them with token volumes disclosed by closed-model companies and hyperscalers. Google provides total token volumes for its platform and API usage in its earnings reports, creating a public reference point for comparison with the open model inference platforms.

The available measurements remain imperfect, but they provide a basis for estimating the scale and growth of the open model inference market.