Perplexity Trains Its Computer Agent on Real Mistakes With Hint-Guided Self-Distillation
4 min
ExplorationBench: Measuring AI Systems' Exploration in Verifiable Alien Worlds
PUBG Ally: A Conversational Embodied Agent as an AI Teammate
Measuring the Serving Stack Instead of the Model: Hidden Confounds in Local Tool-Use Evaluation
Critical-State RL: Diagnosing Trainable States for Multi-Turn Tool Use
Harness-Zero: Harness Distillation via Agent-as-Harness
VideoGen-Agent: Reinforcing Video Generation Agents
CIPL: A Channel-Aware Framework for Recoverable Privacy Leakage in LLM Agents
1
Alibaba Qwen Releases Qwen3.8-Omni-Flash: A 1M-Context Omni-Modal Model Built Around Agentic Audio-Video Understanding and Tool Use
5 min1
ASLEval: Measuring Privacy Exposure Displacement in LLM Agent Sessions
BI-Agent and BI-Bench: Towards Automating End-to-End Business Intelligence
CanvasAnneal: Curriculum Reinforcement Learning for Diffusion Language Models
1
Google Research Releases ToolGrad: Answer-First Framework Hits 99.8% Pass Rate for Tool-Use Data Generation
4 min1
MetroLLM-Bench: Evaluating Language Models as Transit Kiosk Runtimes
1
Procedural Graphs: Self-Evolving Execution Structures for LLM Agents
1
ToolLoop: Closed-Loop Tool-Use Data Synthesis via Decomposed Generation and Dynamic Self-Feedback
1
Procedural Graphs: Self-Evolving Execution Structures for LLM Agents
1
NeoHorse-1: Towards Recursive Self-Improvement via Agentic Post-Training with Routing Harness
1
Agentic Visual Generation: From Generative Models to Agentic Control
Occamy-1.0: Open Pareto-frontier 35B Intelligence for Co-work
An AI CAPTCHA solver talked itself out of the right answer
3 min1
The Evolution of the Agent Harness
9 min