Cognition sells the Fusion harness inside Devin Desktop and CLI, the current product state of the Devin agent. The harness pairs a planning model with a cheaper execution model and is the subject of Cognition's cost-cutting claims.
Fusion is Cognition's harness for driving the Devin agent, and it is now sold inside both Devin Desktop and the Devin CLI. The harness splits an agent run across two models: a planning and review model, and a cheaper execution model. Cognition advises pairing GPT-6 Astra or Anthropic's model as the planner with the SWE-2 coding model as the executor.
The pitch follows from that split. Cognition states Fusion cuts costs by up to 39 percent compared with Codex at comparable scores. The claim comes from the company that makes Devin, not from an independent test, and the 39 percent figure is a ceiling, not a measured average.
Read together, the two points describe the same product position: Cognition is selling harness design as the lever that decides the cost of an agent run. Model choice still matters, since Fusion names specific planners and a specific executor, but the company's argument is that the harness around them decides what a run costs. The harness is available inside the Devin Desktop and Devin CLI products as of the release.
Written with AI from the cited Yomimono stories below, each itself sourced. See our editorial policy. Every fact links the story it came from.
Cognition, the company behind the AI agent Devin, released Fusion, a harness that pairs a planning and review model with a cheaper execution model. Cognition advises GPT-6 Astra or Anthropic's the model as the planning model and the SWE-2 coding model as the executor, and states Fusion cuts costs by up to 39 percent versus Codex at comparable scores.