Three unrelated products and teams named Fusion are active in AI and entertainment: Cognition's Fusion harness for its Devin agent, APOLLO11's apollo11-fusion API, and Stability AI's recently closed Series B round. Only Cognition's Fusion is a harness for the Devin agent; the others use the name for an API endpoint and a funding round.
Yomimono's coverage of the name Fusion is split across three unrelated items. The most recent, from September 14, is Cognition's release of a harness called Fusion that pairs a planning and review model with a cheaper execution model. Cognition advises GPT-6 Astra or Anthropic's the model as the planner, with the SWE-2 coding model as executor, and states Fusion cuts costs by up to 39 percent versus Codex at comparable scores. The harness is sold inside Devin Desktop and the command line.
The two earlier stories use the same name for different things. On August 18, APOLLO11 launched the apollo11-fusion API, a frontier-class LLM endpoint with OpenAI SDK compatibility and Zero Data Retention, offering roughly 3.3 million free tokens for one week and claiming Opus 4.8-level performance at up to 65 percent lower cost, with seven models behind one endpoint. Eleven days later, Stability AI announced a $76 million Series B on August 25, bringing total funding under CEO Prem Akkaraju to $232 million, with new investors including Electronic Arts, Sony Music Group, Universal Music Group, Warner Music Group, AMD Ventures, and Pacific Alliance Ventures.
No story connects the three. Cognition's Fusion is a product; APOLLO11's is an API name; Stability AI's is a financing event. The shared word is the only thread.
Written with AI from the cited Yomimono stories below, each itself sourced. See our editorial policy. Every fact links the story it came from.
Structured graph also available as JSON at /public/entities/fusion.
CC BY 4.0.
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.