One card per story, grouped by what kind of story it is. Sourcing quality runs across the top of every card.
Alpha stakers' share fell from 4.2% to 0.1%, and the token rose 186% the same day (@subnetradarcom). The subnet has no name on chain.
Leadpoet will find AI companies facing rising GPU costs, and their technical buyers, for Pareton's inference-optimisation network (@LeadpoetAI).
Founder Florian says Decision Models is scaling GPU capacity to serve it (@airesearch12). No name or figure.
Two thirds of maker fills now come from open-market flow, up from one third (@mvtrx_79).
Miners' FP8 copy sits closer to the original than Qwen's release; next are 2-bit and larger GLM models (@const_reborn).
It picks the model, provider and reasoning effort for each request (@bitmind).
Its comment letter asks for a disclosure-based framework for issuers that keep trading their own tokens, which covers subnet buybacks (@YumaGroup).
It also merges coldkey swaps, changes Yuma payouts and adds proxies (@subnetradarcom).