# Cognition — Introducing Fusion in Devin Desktop & CLI

- Company: Cognition (cognition.com)
- Announced: 2026-09-11
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://cognition.com/blog/local-fusion
- Record: https://forck.live/items/10255-introducing-fusion-in-devin-desktop-cli
- Subject: Devin / SWE
- Models affected: Fable 5.1, SWE-2, GPT-6 Astra, Fable 5, Opus 4.8, GPT-5.6 Luna
- Pricing: Fable nominally costs twice as much per token

Cognition released Fusion, a model harness that pairs two models — a frontier lead and a cost-effective sidekick — to reduce cost while maintaining performance. The company recommends pairing Fable 5.1 as the lead with SWE-2 as the sidekick. Fusion is now available in Devin Desktop and CLI.

## Evidence

Verbatim from https://cognition.com/blog/local-fusion:

> Fusion is the most efficient frontier harness for Fable and Astra, up to 39% more efficient compared to other model harnesses across major coding benchmarks. Today, we’re making it available in Devin Desktop and CLI.

---

Record: https://forck.live/items/10255-introducing-fusion-in-devin-desktop-cli
Catalogue: https://forck.live/llms.txt
Current issue: https://forck.live/feed.md