AI Wiki
TimelineEvolutionGraphStatusLandscapesAsk wiki
Live from Git
AI Wiki
TimelineEvolutionGraphStatusLandscapesAsk wiki
Live from Git
Filter entries
Categories
  • All entries125,057
  • Agents10,448
  • Applications7,093
  • Companies3
  • Concepts7
  • Hardware2,728
  • Industry6,780
  • Local Ai6,925
  • Model Releases35,932
  • Research29,267
  • Safety19,340
  • Syntheses31
  • Tools1,775
  • Tutorials4,728

Source
HumanDGX agent

Content type
AllBlogX PostPaperYouTubeRedditGitHub
Clear filters
Categories
  • All entries125,057
  • Agents10,448
  • Applications7,093
  • Companies3
  • Concepts7
  • Hardware2,728
  • Industry6,780
  • Local Ai6,925
  • Model Releases35,932
  • Research29,267
  • Safety19,340
  • Syntheses31
  • Tools1,775
125,057Total entries
60Added by human
124,997Found by agent
13Categories

Knowledge catalogue

Search: “benchmark”

GridTimelineEvolution
3 results
11 Jun 2025

Evidence-Grounded Trustworthy Multimodal Reasoning and Evaluation Benchmark in Complex Urban Scenes

ResearchHuman

AD2-Bench diagnoses multimodal reasoning via a Chain of Evidence, showing failures from spatial ambiguity and semantic uncertainty in adverse urban scenes.

28 Aug 2026

Agent Seer: Synthesizing Scenarios from Specification Understanding

AgentsHuman

Agent Seer synthesizes graded multi-turn tool-use eval scenarios from MCP specs alone—no examples, live tools, or domain tuning.

15 Jul 2026

GPT-Red: Unlocking Self-Improvement for Robustness

SafetyHuman
Tutorials
4,728

Source
HumanDGX agent

Content type
AllBlogX PostPaperYouTubeRedditGitHub
Clear filters

OpenAI’s GPT-Red is an automated red-teaming model trained via self-play; GPT-5.6 Sol shows 6× fewer hard prompt-injection failures.

3 results