MiroThinker is a deep research agent optimized for complex research and prediction tasks. Our latest models, MiroThinker-1.7 and MiroThinker-H1, achieve 74.0 and 88.2 on the BrowseComp, respectively.
🏆 Top-1 on 5+ benchmarks | Web UI | Supports MiroThinker, Claude, Kimi, OpenAI
[NeurIPS 2025] 🌐 WebThinker: Empowering Large Reasoning Models with Deep Research Capability
An Go library of synchronization primitives to help make use of hardware transactional memory (HTM)
An easy-to-use evaluation tool for running Humanity's Last Exam on (locally) hosted Ollama instances.
Tiny experiments that evaluate how different social-cue prompts influence large language model (LLM) decision-making on HLE dataset multiple-choice questions.
HyDRA is a hierarchical deep research agent framework that orchestrates planning, tool usage, web exploration, and synthesis through robust context management, enabling reliable multi-step reasoning under strict output and token constraints.
This repository explores how social cues impact decision-making in large language models using a structured pipeline. Join us in analyzing various prompts and their effects on responses to multiple-choice questions! 🐙✨
🎨 Elevate image editing with EditThinker, enabling iterative reasoning for improved instruction adherence and enhanced visual quality.
🧠 Transform ideas into structured insights with MiroThinker, a powerful tool for enhancing creativity and collaboration in your projects.
Add a description, image, and links to the hle topic page so that developers can more easily learn about it.
To associate your repository with the hle topic, visit your repo's landing page and select "manage topics."