NEAR Protocol ($NEAR) co-founder Illia Polosukhin has launched Delta League, a public competition that invites anyone to submit improvements to an open artificial intelligence model. The platform was announced by Polosukhin (@ilblackdragon) and is powered by @near_ai for the @NEARProtocol ecosystem.

How Delta League Works

Participants submit edits to a model's weights — the numbers inside the model that shape its answers. According to the Delta League website, entrants can submit a LoRA, a sparse or dense weight difference, or a full checkpoint, and every change preserves its model lineage. LoRAs — low-rank adaptations — are a compact fine-tuning technique that trains only a small slice of a model's parameters, rather than retraining the whole model. The platform verifies the submitted weights and then runs validation and private tests, so that entrants can compare their scores with those of the original model.

Each edit is scored on a hidden test set, and every verified version stays open for others to build on. Holding test data back is a standard evaluation safeguard in machine learning: it measures performance on examples the contributor has never seen. This structure lets contributors extend earlier work rather than start from scratch, with each change recorded against the model it modified.

First Test on Qwen3.5 0.8B

The first test uses Qwen3.5 0.8B, a small open model — the 0.8B label refers to roughly 800 million parameters. Five verified edits have so far lifted its score from 18.6% to 19%, a gain of 0.4 percentage points. The gain is modest, but it shows the verification process working on a live model.

Qwen3.5 is a family of open-weight models from Alibaba. According to Wikipedia's entry on Qwen, Alibaba released the open-weights Qwen3.5 in February 2026.

What Comes Next

Support for changing model designs and custom tests for businesses will follow in the coming days. Allowing changes to a model's architecture, and not only its weights, would widen the range of contributions the contest can accept. Custom tests would let companies measure edits against their own tasks. Those additions, and any further movement in the base score, are the near-term markers to watch.

The launch fits Polosukhin's wider work on open and user-owned AI. He is a co-author of the 2017 paper "Attention Is All You Need", which introduced the Transformer architecture, and now leads work at NEAR AI, according to Fast Company. In a podcast interview summarised by Open Commons, he argued that releasing model weights alone does not make an AI system open source, and that real openness also requires access to training data, recipes, and methodology. That distinction runs through a wider industry debate, in which several leading model families ship as open weights while their training data stays closed.

Delta League tests a different route to improvement: crowdsourced, verified changes to a shared model, with results scored on data entrants cannot see. The format echoes established machine-learning competitions such as the 2009 Netflix Prize and Kaggle's contests, where distributed entrants improved on shared benchmarks. How far that approach can lift larger models remains to be shown, and the first results are small.