anomalyco / anomalyco/opencode

Vercel AI Gateway `only` provider filter is ignored

Open
#49,521 0 comments 0 reactions 1 assignee View on GitHub

@rekram1-node is already working on this.

Since Sep 17, 2026.

Dominant language
TypeScript
Stars
209k
Forks
27.5k
PR merge metrics
PR metrics pending

Description

Description

Vercel AI Gateway provider filtering still appears to be ignored in OpenCode CLI 1.18.30 + Desktop 1.18.31.

This is the same behavior previously reported in #21130, which was automatically closed after 60 days of inactivity. The original feature request was #2153.

With this config:

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "vercel": {
      "models": {
        "zai/glm-5.3-flash": {
          "options": {
            "only": ["particle"]
          }
        }
      }
    }
  }
}

However, Vercel AI Gateway shows that requests are actually served by other providers. I observed Baseten and Fireworks instead of Particle.

OpenCode logs confirm both requests use the expected Vercel model:

2026-09-17T09:02:43
providerID=vercel modelID=zai/glm-5.3-flash small=true agent=title

2026-09-17T09:02:44
providerID=vercel modelID=zai/glm-5.3-flash small=false agent=build

The Vercel AI Gateway dashboard nevertheless shows these requests
being served by Baseten and Fireworks despite only: ["particle"].

Expected behavior: the request should either use Particle or fail if Particle is unavailable. It should not fall back to another provider.

The current OpenCode Vercel documentation lists only as a supported routing option.

Plugins

No response

OpenCode version

1.18.30

Steps to reproduce

Add the following to the OpenCode config:

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "vercel": {
      "models": {
        "zai/glm-5.3-flash": {
          "options": {
            "only": ["particle"]
          }
        }
      }
    }
  }
}

Verify the effective config:

opencode debug config

It shows:

"only": ["particle"]

Restart OpenCode.

Run:

opencode run -m vercel/zai/glm-5.3-flash "Reply only OK"

Inspect the resulting request in Vercel AI Gateway.

The serving provider is not Particle. In my tests the requests were routed to Baseten and Fireworks.

I reproduced the same behavior with OpenCode Desktop and the CLI.

Screenshot and/or share link
Image
Operating System

macOS 15.7.9

Terminal

Terminal.app

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.