Allora Labs Introduces Forge for Model Competition and Continuous Improvement

Allora Labs has introduced Forge, described as the world's first arena for predictive intelligence, announced in a press release. The platform allows machine learning models to compete on real-world prediction problems while improving through continuous exposure to live data.

Developers can bring a model and a predict() function to Forge, which manages data, infrastructure, and execution. This setup keeps the focus on model performance rather than operational tasks. Models remain under developer ownership regardless of where they run.

More than 140 partners across the Allora Network use predictions generated by Forge models in production. As a result, developers earn ongoing rewards based on the utility and improvement of their models within the network. CEO Nick Emmons stated that Forge is built on the idea that progress in predictive intelligence will come from multiple competing models that collectively raise accuracy and performance.