June 26 – July 2: 12 frontier papers out of 2,155 read.
Papers rated as moving the field
How we choose“The strongest and the most cost effective agent, Codex GPT-5.5, achieves only approximately 42% success rate.”
The authors, in the abstract
A Unified Benchmark Suite of Realistic Agentic Healthcare Environments for Challenging Frontier AI Agents
“We report an autonomous system that runs this loop with no human in the loop, post-training a 30B Nemotron across four rounds over multiple weeks.”
“Across a suite of frontier and open-weight models, we find that strong performance on single-turn SWE tasks does not reliably transfer to multi-turn, user-driven workflows: the best-performing models solve roughly 50% of single-turn baseline tasks but only 25% of the corresponding SWE-Interact tasks.”
“Human-as-Humanoid yields a 4.8--7.2x raw demonstration-throughput gain over humanoid teleoperation in our data-collection analysis, and on several downstream tasks, policies post-trained only with the converted human labels generalize to real-robot deployment without target-task robot demonstrations.”
“Our extensive online inference experiments reveal a stark reality: state-of-the-art MLLMs struggle significantly with concurrent streams, achieving only about 50% score and exhibiting poor proactive ability.”
“Experiments reveal a substantial reliability gap: agents that perform well with reliable tools often fail under recoverable hazards.”
“Experiments show the scalability of the proposed paradigm and verify that stronger world latent enables stronger downstream readouts.”
Jun 26 – Jul 2: 158 cleared, 12 frontier
The week’s papers have not been sorted into fields yet.
Papers this week that name the model
Papers this week with an author at the lab
146 papers