Opencode vs Deepseek harness: weak starts but relentless execution compared
GodComplecs · reddit · 2026-09-02
A developer compared Qwen 3.8 27b under Opencode vs the Deepseek harness:
- Deepseek harness: excellent one-shot, relentlessly pushes tasks forward with working todos, but gets confused after the initial prompt — button functionality is a pain, and even a detailed spec needs heavy follow-up.
- Opencode: stronger initial comprehension, task continuation feels "like magic", but lacks Deepseek's tenacity.
- Hermes didn't fit his complex tasks (even top closed models err); he asks whether Pi is worth trying.
More from coding & agent
- Structured generation and speculative decoding implementation is complex — remilouf · 2026-09-02
- How to evaluate the quality of an agent interface built on CLI/MCP? — nguyenfamjj · 2026-09-02
- Building the Missing Layer Between AI Spend and ROI — AccBalanced · 2026-09-02
- Four agents, four truths: building a 'second brain' to align company context — max_gladysh · 2026-09-02
- Parallel coding agents tested: token bloat and coordination overhead — TheJoyfulTater · 2026-09-02
- How are you testing AI agents for security before deployment? — Negative_Cookie563 · 2026-09-02