Technology ❯ Machine Learning

Model Evaluation

Performance Metrics Benchmarking Performance Comparison Quality Assurance Open-World Classification Performance Analysis Task-Specific Models Safety Standards Bias in AI Reliability

Anthropic Finds Claude’s Responses Vary by Model Version and Language

Anthropic says modest, structured shifts in Claude’s outputs could change how automated agents behave, according to its behavioral analysis.

Chinese AI Models Collapse Per‑Token Costs, Forcing Developers to Rearchitect How They Use LLMs

Microsoft Details 'Whisper Leak' That Reveals AI Chat Topics, With Some Providers Still Unpatched

Microsoft Open-Sources ‘Magentic Marketplace’ After Tests Show AI Agents Are Easily Manipulated

Enterprises Scale RAG in Production as New Studies Expose Recall Gaps and Offer Targeted Fixes

New Studies Advance Reliability‑Aware, Graph‑Based and Monitored RAG as Real‑World Test Flags Failures

Local RAG Moves Into Practice With Ollama Guides as Analysts Detail Production Risks

AI Reasoning Called into Question After Puzzle Tests and Methods Debate

Allen Institute for AI Launches Molmo: Open-Source Multimodal Models Rivaling Tech Giants