PR #28375 real runtime verification, 2026-11-03
main: e1e520c93c; PR: 278a992f209a389fc7abccad391bfd5a9d731329
Chromium: /usr/bin/google-chrome, Playwright 1.60.0; isolated dev homes; no mocks or fixtures.

main OpenCode v2.0.28 / opencode/beam-free, submitted through web UI:
[
  {
    "status": "failed",
    "failure": {
      "class": "usage_limit",
      "message": "Free usage limit reached. Free models work in OpenCode (https://opencode.ai/); subscribe to OpenCode Go to keep going: https://opencode.ai/go",
      "code": "provider.quota",
      "retryable": null
    },
    "retry": null
  }
]
running retry rows: 0
Codex gpt-6.1-sol: read-only package.json turn completed in web UI.

PR OpenCode v2.0.28 / opencode/beam-free, submitted through web UI:
[
  {
    "status": "failed",
    "failure": {
      "class": "usage_limit",
      "message": "Free usage limit reached. Free models work in OpenCode (https://opencode.ai/); subscribe to OpenCode Go to keep going: https://opencode.ai/go",
      "code": "provider.quota",
      "retryable": null
    },
    "retry": null
  }
]
running retry rows: 0
Codex gpt-6.1-sol: read-only package.json turn completed in web UI.

PR: switched failed OpenCode thread to Codex gpt-6.1-sol; context handoff and hello response completed.
UNVERIFIED: running OpenCode 2 rate-limit retry class, retry recovery, real HTTP 402, native mobile/desktop.
The real free service returned a terminal provider.quota error; it did not schedule retries.