- Score 83
- Official
SpecLook: Speculating Where to Look for Fast and Reliable Agentic MLLMs
arXiv:2611.23483v1 Announce Type: new Abstract: Agentic multimodal large language models (MLLMs) that think with images interleave reasoning with zoom-in tool calls. However, this sequential loop is slow, and its answers carry no reliability score. Prior speculation of tokens, reasoning steps, or answers keeps the loop, and answer speculation also accepts confident errors. We find that the loop mo
So what
What this event means by reading role—not a longer recap.
- BuilderA training-free speculate-and-verify wrapper could cut zoom-in latency for image-reasoning agents and expose per-answer reliability scores without labels.
- ResearcherReframes speculation from answers to verifiable observations, reporting 87.2%/86.2% mean accuracy with 7.7x/4.7x speedups on V*, HR-Bench, and POPE.
Score dimensions
Higher total means read first. Each bar is one factor we use to rank the system pool. How we score