OpenCode bug: reasoning-only model turns end agent silently with no text or tool call
juancarlosm · ghdev · 2026-10-14
A reproduced issue in anomalyco/opencode (#55378) reports a subtle agent stall.
- In agentic sessions with Gemini models (e.g., gemini-3.5-flash-lite), a step after a tool result occasionally streams back only reasoning tokens (e.g., 24 reasoning tokens), with no text, no tool calls, and finishreason "stop".
- OpenCode 2.0.29 settles that turn as a normal completion: the agent stops and waits for the user with no message or error.
- This recurs from issue #49673 (same symptom with DeepSeek V4.1 Flash), which was closed against OpenCode 1 and never reopened.
- Expected behavior: a step with no text and no tool call should be retried or surfaced as an error so the user knows to continue.
More from coding & agent
- Claude Code Pushes Urgent Security Update Across All Installations — umeshonai · 2026-10-23
- Claude cut a 2.6-hour video export to 20 minutes — but only after being told to rebuild the server — byhewar · 2026-10-23
- 3 flight-booking rules my AI agent checks that I'd never do by hand — Efistoffeles · 2026-10-23
- OpenJet claims Qwen3.8-27B coding runs 1.9x faster than llama.cpp on one RTX 3090 — JeremyCMorgan · 2026-10-23
- Building an autonomous SEO agent in Hermes with three skills: programmatic pages, Keywords Everywhere, DataForSEO — EXM7777 · 2026-10-23
- Building a sports betting AI agent with onchain execution and human approval by default: architecture and lessons — Dry-Comfortable-2678 · 2026-10-23