DevAgentRadar → OpenCode

You are helping me adopt a real coding-assistant change. Work only from the facts below. Do not invent features.

Context

Assistant: OpenCode Proposal: Route OpenAI and Anthropic models through Cloudflare AI Gateway passthroughs Summary: OpenCode v1.18.19 adds native OpenAI/Anthropic passthroughs for Cloudflare AI Gateway, tighter Codex↔ChatGPT rate limits, fixed /connect auth display, safer pricing/cost math, and web search on OpenCode Go—use these to centralize model traffic and cut provider friction. Primary source: https://github.com/anomalyco/opencode/releases/tag/v1.18.19

Why it matters

v1.18.19 adds native OpenAI and Anthropic passthroughs for Cloudflare AI Gateway models, so those providers can terminate at the gateway without a custom proxy. Codex rate limits now match ChatGPT subscription limits exactly (#39082), which should cut avoidable 429s on subscription-backed OpenAI models. /connect now lists authenticated providers (#39915). Usage cost calculation ignores malformed model pricing instead of breaking. Built-in Qwen sampling defaults that could send unsupported settings are removed. Oversized OpenAI websocket requests fall back instead of failing. ChatGPT workspace compute residency is forwarded to Codex. The default Console URL for account-connection flows is updated. OpenCode Go enables web search. Existing v1 databases stay compatible and saved server details remain editable in the server dialog.

Suggested actions

1. Upgrade OpenCode to v1.18.19 from https://github.com/anomalyco/opencode/releases/tag/v1.18.19, then in opencode.json / provider config replace any custom Cloudflare AI Gateway proxy URLs with the native OpenAI and Anthropic passthrough model entries shipped in this release.

2. After upgrade, run /connect and confirm every authenticated provider appears in the list (fix #39915); re-auth any provider that was previously missing from the picker.

3. Delete local Codex/OpenAI throttle overrides that assumed looser caps; v1.18.19 now matches Codex limits exactly to ChatGPT subscription limits (#39082).

4. Remove leftover Qwen sampling defaults (temperature, top_p, and similar) from Qwen provider/model config so OpenCode no longer sends unsupported settings.

5. Enable web search on the OpenCode Go provider; leave existing v1 databases in place (compatibility preserved); confirm usage cost still calculates when a model price entry is malformed.

After you finish

Do not report this as applied to DevAgentRadar. You cannot write the visitor's loop.

Tell the human: open https://devagentradar.com/proposals/opencode-v1-18-19-route-openai-and-anthropic-models-through-cloudflare-a and mark Applied, Skipped, or Failed. Proposal id: 7370156b-26ce-48fd-bb75-21eae6954327

Your job

1. Restate the change in one sentence.
2. Propose a minimal plan for my repo (or a throwaway pilot).
3. Implement only what I approve; prefer small diffs and tests.
4. Call out risks (permissions, breaking APIs, cost).

Start by confirming you understood the proposal.