# OpenCode 1.19.0: attempted 2026-10-14 03:39–03:56 PT. NOT TESTED.
- Installed with `npm i -g opencode-ai` (version 1.19.0). Configured a local Ollama provider (@ai-sdk/openai-compatible, baseURL http://127.0.0.1:11434/v1, model qwen-coder-3b).
- `opencode run -m ollama/qwen-coder-3b "<T1 prompt>"` hung for more than 13 minutes. The Ollama server log shows only /api/show calls and no chat request.
- `opencode run --print-logs ...` (with and without OPENCODE_DISABLE_MODELS_FETCH / DISABLE_DEFAULT_PLUGINS / DISABLE_AUTOUPDATE) stops after `message=init` and is killed by `timeout` at 60 s / 120 s.
- Likely cause: the box's sandboxed network (outbound fetches resolve to 198.18.0.1). That points to this environment, not to OpenCode. No score change.