Skip to content

MnemosyneC v0.9.9

Choose a tag to compare

View all tags
@Upekrithen Upekrithen released this 27 Jul 00:30
v0.9.9
fc2464e

MnemosyneC v0.9.9

WAN responder honors each peer's selected model. The mesh responder now answers with the peer's actually-selected local model (via activeBrain / brain_registry) instead of a hardcoded gemma4:12b. This fixes 32GB peers set to gemma2:9b silently running 12b, splitting VRAM, timing out, and abstaining on WAN-dispatched questions -- and it unlocks per-peer model diversity across the cooperative mesh. peer_presence now advertises each peer's real answer model, and the sku-upgrade tier map is de-hardcoded to hardware-aware RAM tiers (2b / 9b / 12b).

Also carries v0.9.8 CleanShell: the lean shell is now the default front door -- six real tabs plus combined Substrate, dead surfaces un-routed, stubs cut.

  • SHA-256: 3e4156867c250bb90205277734899a75389ab883d48029672b5cf6bdb2415941
  • Size: 483,600,158 bytes (~461.2 MB)

A participation-based cooperative AI peer by Liana Banyan Corporation. Platform credits are not securities. SSPL.

Assets 3
Loading