Taming Anthropic's Opus 5: Brilliant in Flashes, Frustrating in Practice
every · x · 2026-07-29
The Every team shares their in-depth experience testing Anthropic's latest model, Claude Opus 5. They found the model to be brilliant in flashes but frustrating in practice, requiring too much management and repeated prompting to keep responses simple.
They discovered a workflow to "tame" Opus 5: assign it one substantial job with a clear finish line, then leave it alone. The article also discusses how to audit agent instructions built for older models so they don't get in the new one's way, and theorizes why one-shot AI demos of video games clog up social feeds.
Related event: Claude Opus 5 Official Prompting Guide and Practical Tips(7 posts)→
More from coding & agent
- Harness open-sources desktop and mobile apps for orchestrating agent fleets across frameworks and machines — dee_hw · 2026-10-16
- Open-Source Google Analytics MCP Server Lets LLMs Query GA Data via Natural Language — modelcontextprotocol · 2026-10-16
- LLM ideation experiment: just ask for non-obvious ideas — and Opus beats smaller models by a wide margin — arjunrajlab · 2026-10-16
- a16z podcast: AI cut a 6-month feature to 10 days and widened engineer gaps 10x — a16z · 2026-10-16
- "Hold On, My Agents Are Freezing My Screen": The Daily Grind of the Agent Era — every · 2026-10-16
- Dev builds MCP feeding Claude 3,000 real launch videos to fix samey AI output — Designer_Landscape25 · 2026-10-16