# Engineers... STOP Picking GPT-5.6 Sol OR Claude Fable 5… FUSE THEM

## Executive summary

The video argues that in agentic engineering, the optimal approach is not to choose a single 'winner' model (e.g., GPT 5.6 Sol vs. Claude Fable 5), but rather to implement Model Fusion. This involves building custom agent harnesses that coordinate multiple state-of-the-art models working together. The process utilizes specialized commands—`/opinion` for diverse perspectives, `/fusion` for consolidating results, and `/auto validate` for intelligent on-the-fly review—to significantly outperform single-agent workflows.

## Key takeaways

- Model Fusion: AND, Not OR: The most powerful approach is combining the compute and intelligence of multiple models rather than selecting a single winner. This pattern combines concepts previously known as architect editor, prompt chaining, and agent chaining.
- Three Core Commands for Orchestration: A custom fusion harness uses three commands: `/opinion` (to gather multiple perspectives), `/fusion` (to combine and consolidate results), and `/auto validate` (for intelligent on-the-fly validation, addressing the review constraint of agentic engineering).
- Value of Fusion vs. Single Agent: Fusion allows agents to identify consensus, divergence, and discarded information, providing a comprehensive view that is critical for high-stakes strategic decisions.

## Technical details

- Agentic Workflow Components: The fusion harness orchestrates a coordinated two-agent team. The process involves: 1) Running parallel agents (e.g., Claude Sonnet 5 and GPT 5.6 Terra) to gather raw data/opinions. 2) Using `/fusion` to consolidate these results into a single, comprehensive output. 3) Implementing `/auto validate`, which runs validation scripts before execution, ensuring the work is verifiable.
- Benchmarking and Performance: The video compares models like GPT 5.6 Terra and Claude Sonnet 5 on simple tasks, noting that while one might be faster (e.g., Terra in 4.5 seconds), the key metric is the ability to run agents side-by-side for comparison. The final benchmark showed a combined approach achieving significant speedups (e.g., up to 10,000X) over native SQL light autocommits.
- Advanced Agentic Concepts: The concept of 'micro SDLCs' (Software Development Life Cycles) is built into the harness. The system emphasizes that owning and customizing the agent harness (e.g., using a custom PI coding agent) is crucial for overcoming external tool limitations.

## Practical implications

- Build engineers should focus on owning and customizing their agent harnesses rather than relying solely on out-of-the-box tools.
- The workflow of Opinion -> Fusion -> Auto Validate provides a robust, multi-layered approach to complex problem-solving in AI development.
- Focusing on 'harness engineering' (prompt, context, and harness design) is the highest leverage point for maximizing agentic results.

## Topics

Agentic Engineering, Model Fusion, Prompt Engineering, Software Development Life Cycle (SDLC), AI Architecture, Get The Pi Fusion Harness, FORGET LOOP ENGINEERING, Pi Coding Agent, Devin Fusion, Open Router Fusion, Aider Architect Editor (original fusion)

Source: https://www.youtube.com/watch?v=AQl5Q-0l7FQ