Perplexity’s JOLT method matches GRPO accuracy with 6.5x fewer training tokens

A new technique from Perplexity Research and NVIDIA lets a single model act as both teacher and student during training Training a large language model to reason well is expensive. Perplexity Research thinks it has found a way to make that process considerably cheaper. The company, working with NVIDIA , introduced a method called JOLT, short for Joint On-Policy Learning and Teaching, on October 21, 2026. On the MATH benchmark, JOLT reached the same accuracy as Group Relative Policy Optimization (GRPO) while using 6.5 times fewer training tokens . That matters because GRPO is already considered one of the more efficient options in the field. A single model plays two roles at once: it teaches, and it learns.

BeCrypto News shows a short excerpt and links to the publisher. The full report belongs to Crypto Briefing.