Three LLM rerankers each uniquely solve different queries: Terra 28.3, Luna 16.0
CShorten30 · x · 2026-11-16
Tweet 11/17: evidence for the 'width' lever at the listwise stage — averaged over 3 trials, GPT-5.6 Terra uniquely solves 28.3 queries, Luna 16.0, and GPT-5.4 Mini 10.3, showing the models fail on largely disjoint query sets, the precondition for any fusion experiment.
Related event: Reranking Study Finds Two-Stage Pipeline Beats Fusion and Deeper Windows(13 posts)→
More from Research
- BeyondBinding challenge launches: AI-designed antibodies get wet-lab tested — QuanquanGu · 2026-11-17
- Singularity-Bench: Opus 5.5 with Claude Code beats human baselines on building AI systems — armand_ruiz · 2026-11-17
- Open Vision Banana: fully open recipe turns image generators into multi-task perception systems — NielsRogge · 2026-11-17
- WebArena-Pro debuts as largest browser-agent benchmark, top agents score 62.3% — xhluca · 2026-11-17
- NeurIPS 2026 paper: RL-trained skills transfer poorly to composed reasoning tasks — StanfordAILab · 2026-11-17
- OpenAI warns CoT monitorability will degrade as models scale, researchers push back — gleech · 2026-11-17