For failing queries, your evaluation shows the gold chunk is present in the top-5 context 92% of the time. The answers still ignore it or hedge. The team proposes migrating to a different vector database. Where should effort go?

1. A Swap to a different vector database, since a better ANN index will raise recall at 5 above its current level.
2. B Increase top-k from 5 to 25 so the generator sees more supporting evidence for each failing question.
3. C Add a cross-encoder reranker, because the gold chunk is likely ranked too low within the top-5 window.
4. D Treat it as a generation problem and iterate on the prompt, context formatting, and grounding instructions.