Same Model, Different Harness: OpenCode Refuses Far More Than Codex in Side-by-Side Tests
Current_Balance6692 · reddit · 2026-10-12
A Reddit user compared OpenCode against Codex's official harness and found a stark difference: the same models that comply in Codex kept intervening in OpenCode, rejecting requests and adding safety caveats — especially around bio and nuclear topics. Side-by-side runs ruled out variance, and the pattern held across model swaps.
Models tested at highest thinking: Sol 6.1, Astra, GLM 5.3, Muse Spark 1.3, and Qwen 3.8 Max. Takeaway: the harness layer's safety policy dramatically shapes real-world usability — choosing a coding tool means evaluating the harness, not just the model.
More from coding & agent
- Predicting a 75% chance Anthropic or OpenAI ships an LLM-optimized programming language this year — dh7net · 2026-10-12
- 'Software is dead' takes get the camera analogy rebuttal: it's software's Kodak Brownie moment — mgill25 · 2026-10-12
- CUT, a Timeline Editor for MiniMax H3 in ComfyUI, Now Free in Original Version — rileygstaliger · 2026-10-12
- What breaks first when your AI agent runs unattended? Not the model — RehanShakir · 2026-10-12
- Eight AI agents, one media-buying benchmark: two ace 12/12, two fail 0/12 — ML_Felix · 2026-10-12
- pi author: agent self-modification is really user-initiated, agents don't need to rewrite their harness — lucasmeijer · 2026-10-12