Lightning Weave improves reasoning accuracy and efficiency in AI models
Lightning Weave: Improving the Accuracy-Efficiency Frontier of Reasoning Models through Capability Composition
Artificial Intelligence
Summary
Improving both accuracy and speed in AI reasoning is hard because they often conflict. The paper presents Lightning Weave, a method that combines different specialized AI models into one that balances accuracy and efficiency better. It does this by learning from the differences in specialized models using a new training technique. Experiments show this approach raises accuracy while reducing the number of tokens needed for reasoning on math and coding tasks.
What this means in practice
- •For ai model engineers: Build more efficient AI reasoning systems that improve accuracy without increasing response time by combining specialist capabilities in one model.
- •For automated math and code assistants: Improve automated tools that solve math problems and write code by applying Lightning Weave to balance solution accuracy and computational cost.
Authors
Yecheng Wu, Song Han, Han Cai
Abstract
A core goal of efficient reasoning is to improve the accuracy-efficiency frontier. However, jointly improving reasoning accuracy and inference efficiency can be challenging, as the two objectives can favor different reasoning behaviors. Independently post-trained models already offer distinct strengths in accuracy and efficiency. We introduce Lightning Weave, a post-training framework that extracts and composes these independently learned capabilities in a single student through on-policy distillation. Each acquired capability is represented by the policy shift from the model before post-training to the resulting specialist. Lightning Weave combines aligned log-ratio shifts at shared student token states and uses Tilted-Target DOPD to convert the cached signals into a stable learning target. Each anchor pair scores the cached trajectories once, enabling subsequent student training without serving multiple live anchor models concurrently. Across diverse student models and benchmarks in mathematics and code, Lightning Weave substantially improves upon the base students and achieves a state-of-the-art accuracy-efficiency frontier. On Qwen3.5-4B, it raises HMMT 2025 accuracy from 59.2% to 64.0% with 10.7% fewer response tokens, and LiveCodeBench v5 accuracy from 41.7% to 54.2% with 9.6% fewer response tokens. Adjusting the relative strengths of the anchor signals yields a strong empirical accuracy-efficiency Pareto frontier. These results establish Lightning Weave as a new practical route to efficient reasoning through capability composition. Code will be released soon.