Apex Fusion Shows MIT's AI Agent Societies Do Not Need a Frontier Model to Work

This story was originally published on HackerNoon at: https://hackernoon.com/apex-fusion-shows-mits-ai-agent-societies-do-not-need-a-frontier-model-to-work.
Apex Fusion ran MIT's SwarmWorld experiment unchanged on four RTX 3090 GPUs with an open-weight model: 9 of 12 effects held and the society reached 16 of 16.
Check more stories related to undefined at: https://hackernoon.com/c/undefined. You can also check exclusive content about #apex-fusion, #ai-agents, #ai, #technology, #web3, #good-company, #blockchain, #ai-and-ml, and more.

This story was written by: @ishanpandey. Learn more about this writer by checking @ishanpandey's about page, and for more stories, please visit hackernoon.com.

Apex Fusion Foundation reran MIT's SwarmWorld experiment with 50 agents on four consumer RTX 3090 GPUs in about 26 hours and 12,800 model calls, with MIT's code unchanged. Nine of twelve paired effects matched MIT's direction, including every portfolio-resilience gain, which was larger in the replication. A single Qwen3.8-27B agent reached 6 of 16 research milestones against 7 to 10 for MIT's model, while the society reached 16 of 16 against 14. A hidden KV-cache precision setting changed results more than the gap between a 27B and a 235B model. The swarm also ran on Vector, Apex Fusion's settlement layer. The foundation is also in the process of joining the Agentic AI Foundation.