Opus 5 Review: Technically Brilliant but Stiff, Tailored for Sub-Agents Not Solo Chat
MicahBerkley · x · 2026-07-28
The author shares deep insights into using Anthropic's Opus 5, noting that while it is technically brilliant—with next-level error surfacing and visual taste—its conversational flow and creative ideation feel incredibly stiff.
Rather than a bug, the author argues this is a deliberate feature for the sub-agent era. Anthropic is training these models for the swarm, acting as technical specialists managed by an orchestrator. When treated as a solo creative lead, it frustrates; but when used as a technical sub-agent reporting to a manager, it delivers 10/10 execution.
More from coding & agent
- Three essays on AI-first software ask how hard contracts survive cheap code — _xjdr · 2026-07-28
- Krea 2 reportedly generated all 50 car models from a single prompt set — poopoo_fingers · 2026-07-28
- A creator curates 19 open-source repositories for building AI agents — MaryamMiradi · 2026-07-28
- Google Cloud adds spend caps and sandboxes for untrusted agent workloads in Cloud Run — steren · 2026-07-28
- Parallax OS: An Open-Source AI-Native Operating System for Autonomous Agents — avlok · 2026-07-28
- Parallax pitches a personal-agent operating system built on LLMs — avlok · 2026-07-28