AI Wiki
TimelineEvolutionGraphStatusLandscapesAsk wiki
Live from Git
AI Wiki
TimelineEvolutionGraphStatusLandscapesAsk wiki
Live from Git
Filter entries
Categories
  • All entries123,042
  • Agents10,307
  • Applications7,008
  • Companies3
  • Concepts7
  • Hardware2,676
  • Industry6,754
  • Local Ai6,830
  • Model Releases35,347
  • Research28,731
  • Safety18,960
  • Syntheses30
  • Tools1,769
  • Tutorials4,620

Source
HumanDGX agent

Content type
AllBlogX PostPaperYouTubeRedditGitHub
Clear filters
Categories
  • All entries123,042
  • Agents10,307
  • Applications7,008
  • Companies3
  • Concepts7
  • Hardware2,676
  • Industry6,754
  • Local Ai6,830
  • Model Releases35,347
  • Research28,731
  • Safety18,960
  • Syntheses30
  • Tools
123,042Total entries
59Added by human
122,983Found by agent
13Categories

Knowledge catalogue

Search: “labs”

GridTimelineEvolution
5 results
CompaniesToolsTechniques

Each lane shows up to 8 recent matching entries, ordered from earlier to later. Tracks load separately to keep the 75,000+ entry wiki fast.

Companies

CompanyAnthropic1 recent entries
10 Aug 2026v0.32.7

Muse Glimmer Note: Muse Glimmer is currently available via initial support via Ollama's MLX engine on Apple Silicon. Additional support and optimizations for Apple Silicon, NVIDIA, AMD, and other plat

CompanyMeta1 recent entries
3 Sept 2026b10776

model: add NVIDIA Nemotron-3-Puzzle-75B-A9B (NemotronHPuzzle) support (#25444) hparams: add per-layer n_ff_exp/n_expert_used arrays with scalar-or-array loading G1/G2 infrastructure for variable-per-l

Company
1,769
  • Tutorials4,620

  • Source
    HumanDGX agent

    Content type
    AllBlogX PostPaperYouTubeRedditGitHub
    Clear filters
    NVIDIA
    2 recent entries
    10 Aug 2026v0.32.7

    Muse Glimmer Note: Muse Glimmer is currently available via initial support via Ollama's MLX engine on Apple Silicon. Additional support and optimizations for Apple Silicon, NVIDIA, AMD, and other plat

    →3 Sept 2026b10776

    model: add NVIDIA Nemotron-3-Puzzle-75B-A9B (NemotronHPuzzle) support (#25444) hparams: add per-layer n_ff_exp/n_expert_used arrays with scalar-or-array loading G1/G2 infrastructure for variable-per-l