The same model can behave very differently because the harness changes the protocol
remilouf · x · 2026-08-04
The article argues that model performance in agent systems depends heavily on the harness, not just the weights.
Core claim
Run the same model through Pi, Codex, Claude Code, or OpenCode, and you can get noticeably different behavior: one runtime plans better, another makes cleaner tool calls, and another gets stuck retrying. The weights are identical, but the outcomes are not.
Why that happens
The author says the usual execution loop is deceptively simple:
- send the conversation and tools to the model
- execute any tool calls
- feed results back
- repeat until done
That loop may be a few hundred lines of code, but the protocol around it changes everything. The piece’s central thesis is that a harness is not just an implementation detail; it is the protocol by which a model interacts with the outside world, and that protocol meaningfully shapes capability.
More from coding & agent
- Long-context voice agents ditch turn detection with async compaction handoff — juberti · 2026-08-04
- Phone-installable PWAs become custom AI agents with Tailscale and Codex — johnlindquist · 2026-08-04
- DeepSeek V4 Flash tops its price tier on GBench and looks much smarter in one-shot tasks — teortaxesTex · 2026-08-04
- Agent DevTools debugs memory, retrieval, and tool calls in local runs — No_Firefighter8428 · 2026-08-04
- Codex edited a full three-camera episode end to end in under 3 hours — danshipper · 2026-08-04
- CTO says an engineer automated 60% of his job with an AI agent and got promoted — sloppenheimer · 2026-08-04