MTRouter learns turn-level model utility predictors from logged trajectories using history-model joint embeddings, delivering 58.7% cost reduction on ScienceWorld and 43.4% on HLE while matching or exceeding GPT-5 performance.
Output thefull original contextas much as possible (it can exceed three paragraphs)
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
MTRouter: Cost-Aware Multi-Turn LLM Routing with History-Model Joint Embeddings
MTRouter learns turn-level model utility predictors from logged trajectories using history-model joint embeddings, delivering 58.7% cost reduction on ScienceWorld and 43.4% on HLE while matching or exceeding GPT-5 performance.