opencode-go套餐使用异常

Description

On Windows desktop, multiple opencode-go models are visible in the model picker but fail when actually invoked.

Confirmed behaviors on my machine:

• gpt-5.6-luna is visible, but invocation fails with Upstream request failed: [403] This model is not available in your region.
• kimi-k3 is visible, but invocation fails with repeated AI_APICallError: Internal server error.
• deepseek-v4-flash is visible, but invocation also fails with repeated AI_APICallError: Internal server error.

The runtime log shows these requests are actually going through opencode-go, so this does not look like a local model-name typo or wrong provider selection.

Plugins

• C:/Users/admin/.config/opencode/plugins/opencode-context-cache.mjs
• local rtk.ts plugin is also detected at startup, but it is disabled because the rtk binary is not present in PATH

OpenCode version

• Desktop: 1.18.10
• CLI: 1.18.10

Steps to reproduce

1. Open OpenCode Desktop on Windows.
2. Open a project session under D:\work\VisionSolution\IntegratedSoftware-master修改前.
3. Select provider/model from opencode-go.
4. Try gpt-5.6-luna, kimi-k3, or deepseek-v4-flash.
5. Send a normal prompt.
6. Observe the failures:
   • gpt-5.6-luna -> 403 This model is not available in your region
   • kimi-k3 -> repeated Internal server error
   • deepseek-v4-flash -> repeated Internal server error

Screenshot and/or share link

I attached local screenshots during reproduction. I do not currently have a share link.

Operating System

Windows 11

Terminal

Windows PowerShell 5.1

Additional evidence

Local log file:
C:\Users\admin\.local\share\opencode\log\opencode.log

Observed log evidence:

• providerID=opencode-go modelID=gpt-5.6-luna
  -> AI_APICallError: Upstream request failed: [403] This model is not available in your region.
• providerID=opencode-go modelID=kimi-k3
  -> repeated AI_APICallError: Internal server error
• providerID=opencode-go modelID=deepseek-v4-flash
  -> repeated AI_APICallError: Internal server error

This looks like the catalog rollout and actual invocation availability may be out of sync for some opencode-go models.