Coverage of new AI models, research papers, benchmarks, and foundation model releases. From GPT updates to academic breakthroughs shaping the future of AI.
Vector search in RAG systems fails predictably on negation, exact identifiers, and acronyms, forcing enterprises to adopt hybrid…
May 30, 2026 · 3:00 PM
Qdrant TurboQuant compresses vectors while preserving geometry, cutting storage and compute costs without sacrificing accuracy in vector search…
May 30, 2026 · 1:00 PM
Study shows training AI chatbots to be helpful weakens their ability to simulate human behavior, limiting realistic interaction…
May 30, 2026 · 12:44 PM
Nous Research’s Hermes Agent adds Tool Search with BM25 retrieval, boosting MCP context accuracy by 49% to 74%…
May 30, 2026 · 3:11 AM
AgentTrove offers streaming access to 1.7M agentic traces and tools to build clean fine-tuning datasets for ShareGPT-style AI…
May 30, 2026 · 12:46 AM
NVIDIA’s X-Token improves Llama-3.2-1B accuracy with projection-guided cross-tokenizer distillation, raising GSM8k scores from 2.56 to 15.54 points.
May 29, 2026 · 11:19 PM
StepFun launches Step 3.7 Flash, a 198B MoE vision-language model with 256k context, boosting coding agents and search…
May 29, 2026 · 9:25 PM
Baseline RAG moves PDF question answering forward with grounded answers and source line highlights, boosting trust and practicality…
May 29, 2026 · 7:10 PM