• attention — The mechanism that lets a model weigh which earlier words matter for the word it's currently processing — the core operation of a transformer.

An open trillion-parameter reasoner aimed at planning and multi-step tool calls, with an attention mix meant to cut the cost of long reasoning traces. Weights are downloadable now.

• postAReaL v0.2 releases 7B and 32B reasoning models with their training data
• postChat2Graph puts a multi-agent system on top of a graph database
• postAwex syncs trillion-parameter weights to inference in under six seconds
• postLing-2.6-1T arrives as Ant's agent-focused open flagship

• postRing-2.6-1T Ships With a Reasoning-Effort Dial for Agent WorkloadsAnt Ling
• postRing-2.6-1T opens up with two reasoning effort levelsAnt Ling
• postLing and Ring 2.6 technical report lands with two open base checkpointsAnt Ling

Checking sign-in…

Loading comments…