Review Rarely Beats the Stronger Model: Separating Verification from Capability Transfer in LLM Pipelines

arXiv:2610.15897v2 Announce Type: cross Abstract: Qwen3-8B reviewing Llama-3.1-8B in math adds 33.2 points, yet Qwen3-8B alone is 6.4 points more accurate than the result. Reviewing another model's answer is usually judged by its gain over the generator. We show that this confuses verification, which adds accuracy beyond either model alone, with capability transfer, where a stronger reviewer supplies answers it could have given on its own. Measured against the stronger model alone, across 48 pairs of 12 open-weight models (3B to 72B) in math, knowledge, code and reasoning, review beats it in only 9 of 96 review directions, and does worse the larger the capability gap. The reason is anchoring: the reviewer mostly follows the generator, discarding its own correct answers. Review costs 2.6 to 2.8 times as much as the stronger model alone, while a selector using the same two calls is 2.7 to 4.2 points more accurate than review and cheaper. Studies of verification should report the stronger model alone, not only the gain over the generator.

Related

• When Does a Second Model Help? Cross-Model Review in LLM Verification
• Detection Without Correction: A Two-Parameter Decomposition of Multi-Stage LLM Pipelines
• How Closely Do LLM Reviews Align with Human Peer Review?

Source: arXiv cs.AI | 2026-10-15