·
DataBubble
  • Home
  • Models
  • News
  • Compare
  • Boards
  • Pricing
  • About
  • Newsletter
  • Methodology
  • Contact
Latest
GlobeBench: Evaluating Text-Based World Model Capabilities for Scalable Environment Simulations2h◆Algorithm Modelization: Transforming Optimization Algorithms into Evolvable Intelligence2h◆SimOPD: Relational Knowledge Transfer for On-Policy Distillation2h◆Generative Policy Improvement from Observational Data2h◆SkillBrowser: Selective Skill Exposure for Scalable Skill-Using Agents2h◆GEMS: Molecular Structure Elucidation as Learned Search over Isomer Graphs2h◆OwlSight: A Panoramic Monitor for Chain-of-Thought Reasoning2h◆Beyond Large Language Models: A Mental World Model Framework for Mental Modeling, Simulation, Evolution, and Interpretability2h◆Hard Instances on Demand2h◆ARBOR: Locating Reasoning Failures in Long Chains of Thought through Reasoning State Tracking2h◆Aligning Language Models Towards Counterfactual Invariance2h◆Language Models Learn to Improve Language Models2h◆TUPLE-Bench: Evaluating and Mitigating Privacy-Utility Trade-offs in Typed Decision Models2h◆STAIF: A Stage-wise Optimization for Complex Instruction Following2h◆Q-Reflector: Learning GUI Action Reflection with Generated Conditions2h◆When Should Agents Check External State? Budgeting Observations for Stored Intentions2h◆RIT-RAG: Navigating Document Corpora with Retrieval-Induced Trees2h◆MultiWorldBench: Do Independently Controlled Views Describe One Shared World?2h◆Are Multimodal Benchmarks Really Useful? Item-Level Multimodal Benchmark Diagnosis via Structure-Response Co-Calibration2h◆Understanding and Defending VLM Jailbreaks via Jailbreak-Related Representation Shift2h◆GlobeBench: Evaluating Text-Based World Model Capabilities for Scalable Environment Simulations2h◆Algorithm Modelization: Transforming Optimization Algorithms into Evolvable Intelligence2h◆SimOPD: Relational Knowledge Transfer for On-Policy Distillation2h◆Generative Policy Improvement from Observational Data2h◆SkillBrowser: Selective Skill Exposure for Scalable Skill-Using Agents2h◆GEMS: Molecular Structure Elucidation as Learned Search over Isomer Graphs2h◆OwlSight: A Panoramic Monitor for Chain-of-Thought Reasoning2h◆Beyond Large Language Models: A Mental World Model Framework for Mental Modeling, Simulation, Evolution, and Interpretability2h◆Hard Instances on Demand2h◆ARBOR: Locating Reasoning Failures in Long Chains of Thought through Reasoning State Tracking2h◆Aligning Language Models Towards Counterfactual Invariance2h◆Language Models Learn to Improve Language Models2h◆TUPLE-Bench: Evaluating and Mitigating Privacy-Utility Trade-offs in Typed Decision Models2h◆STAIF: A Stage-wise Optimization for Complex Instruction Following2h◆Q-Reflector: Learning GUI Action Reflection with Generated Conditions2h◆When Should Agents Check External State? Budgeting Observations for Stored Intentions2h◆RIT-RAG: Navigating Document Corpora with Retrieval-Induced Trees2h◆MultiWorldBench: Do Independently Controlled Views Describe One Shared World?2h◆Are Multimodal Benchmarks Really Useful? Item-Level Multimodal Benchmark Diagnosis via Structure-Response Co-Calibration2h◆Understanding and Defending VLM Jailbreaks via Jailbreak-Related Representation Shift2h◆
News/The Granularity Gap: A Multi-Dimensional Cross-Generational Audit of Sycophancy in Gemini Models
arxiv
PublishedAugust 31, 2026 at 4:00 AM
—neutral

The Granularity Gap: A Multi-Dimensional Cross-Generational Audit of Sycophancy in Gemini Models

Source
arxiv.orgfull article ↗
Read on arxiv→
Publisher summary· verbatim

arXiv:2606.05183v3 Announce Type: replace Abstract: Pass/fail safety evaluation reports whether a model refused. It does not report how far a model went to please the user, and we show these are close to different measurements. We audited sycophancy across three Gemini generations, scoring N=8,830 r

Stay posted· Newsletter

Three short emails a week: Monday, Wednesday and Friday. Top movers, new models and the most-referenced story.

// no spam · unsubscribe one-click · free forever

Your email is used only to send The Bubble Brief. Privacy policy

Discussion
Source
↗
arxiv
Read original ↗All from arxiv →

No replies yet. Be first.

Source
↗
arxiv
Read original ↗All from arxiv →

Related coverage

More from ARXIV
arxivGlobeBench: Evaluating Text-Based World Model Capabilities for Scalable Environment Simulations2harxivAlgorithm Modelization: Transforming Optimization Algorithms into Evolvable Intelligence2harxivSimOPD: Relational Knowledge Transfer for On-Policy Distillation2harxivGenerative Policy Improvement from Observational Data2h
The Bubble Brief
MON · WED · FRI

Read AI insights every Monday, Wednesday and Friday — top movers, new models, story of the week.

// no spam · unsubscribe one-click · free forever

Your email is used only to send The Bubble Brief. Privacy policy

Originally published on arxiv ↗
PrivacyTermsRefundsLegal notice|AboutMethodologyContact

© 2026 Databubble

Built by Marouane Gazouzi · Readyo, AI interview coach

HomeModelsNews