OpenCode

Add PleumRouter as a custom provider in OpenCode, the open-source terminal coding agent.

OpenCode is an open-source coding agent for the terminal (TUI) that can add an OpenAI-compatible endpoint as a custom provider. PleumRouter connects over the OpenAI Chat Completions format (/v1/chat/completions).

Quick start#

1
Install OpenCode
Skip this if it is already installed.
install.sh
macOS and Linux.
```
curl -fsSL https://opencode.ai/install | bash
opencode --version
```
The package is `opencode-ai`.
```
npm install -g opencode-ai
opencode --version
```
Create an API key
In the dashboard's API keys page, create a key that starts with plm_ and copy it. Set daily, weekly or monthly budgets on it while you are there.

Connect to PleumRouter

Pick one of the three. OpenCode reads a custom endpoint only from the provider block in opencode.json, not from environment variables.

pleum launch

Leaves your `opencode.json` alone and applies the provider config to this run only. PleumRouter models are, however, added to OpenCode's model favorites (`model.json`) and stay there.

npm i -g pleumrouter
pleum login              # sign in with the browser
pleum launch opencode    # temporary provider config for this run only

First `export` your key as the `PLEUM_API_KEY` environment variable in your shell. The key is read through `{env:PLEUM_API_KEY}`.

{
  "$schema": "https://opencode.ai/config.json",
  "model": "pleum/gpt-6-sol",
  "provider": {
    "pleum": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "PleumRouter",
      "options": {
        "baseURL": "https://apirouter.pleum.ai/v1",
        "apiKey": "{env:PLEUM_API_KEY}"
      },
      "models": {
        "gpt-6-sol": {},
        "claude-sonnet-5": {},
        "deepseek-ai/DeepSeek-V4.1-Flash": {}
      }
    }
  }
}

`/connect` stores only the key. The address and models still belong in `opencode.json`.

# 1) in OpenCode: run /connect, choose Other, enter provider id "pleum", paste your plm_ key
#    (stored in ~/.local/share/opencode/auth.json)
#
# 2) opencode.json — same block as above, but without "apiKey":
#    provider.pleum.options = { "baseURL": "https://apirouter.pleum.ai/v1" }

4
Pick a model and check
Run opencode and choose a PleumRouter model in /models. If opencode.json has a top-level model set to pleum/<model id>, OpenCode starts with that model. A reply to one prompt means you are connected.
Terminal
```
cd your-project
opencode

# inside OpenCode, pick a PleumRouter model
/models
```

Why use PleumRouter with OpenCode?#

Pay only for what you use from prepaid credits. Credits are deducted separately from the tool maker's subscription.
Set budgets and limits per key. Daily, weekly and monthly budgets, model allow-lists and team budgets help you manage what an agent spends.
One key, many models. Pick Claude, GPT, Gemini or open models; automatic retries and fallback cover provider outages; usage shows up in the dashboard.

How it works#

OpenCode uses the @ai-sdk/openai-compatible adapter to send OpenAI Chat Completions requests to {baseURL}/chat/completions. PleumRouter accepts them, handles streaming (SSE) and tool calls as they are, and forwards them to the provider of the model you chose. Auth sends your plm_ key as Authorization: Bearer. In provider.pleum, pleum is the provider ID, and models are referenced as pleum/<model id>.

Configuration reference#

Setting	Notes
modelpleum/gpt-6-sol	Optional. Default model at startup (`provider id/model id`).
provider.pleum.npm@ai-sdk/openai-compatible	The OpenAI-compatible Chat Completions adapter.
provider.pleum.namePleumRouter	The name shown in the model picker.
options.baseURLhttps://apirouter.pleum.ai/v1	Ends at `/v1`. Do not append `/chat/completions`.
options.apiKey{env:PLEUM_API_KEY}	Reads the key from an environment variable. Omit it if you stored the key with `/connect`.
models{ "gpt-6-sol": {} }	The model IDs to show in `/models`. Only the ones listed here appear.
models.<id>.limit{ "context": 1050000, "output": 128000 }	Optional. Context and output limits, so compaction happens at the right time.
Config file location./opencode.json	Per project. For all projects, use `~/.config/opencode/opencode.json`.
Key storage~/.local/share/opencode/auth.json	Where a key entered with `/connect` is stored.

Choosing models#

Only the IDs you list in models appear in /models. Look up exact IDs in the model catalog or with GET /v1/models. Adding limit.context and limit.output per model lets OpenCode compact at the right time, and name is the label shown in the picker. Add -latest (as in claude-sonnet-latest) to always get the newest release of a family. To skip the picker on startup, set a top-level model to pleum/<model id>. It applies to the terminal, the desktop app and opencode run. OpenCode 2 does not accept opencode --model, so this setting decides the starting model, while opencode run accepts --model pleum/<model id>.

opencode.json

{
  "provider": {
    "pleum": {
      "models": {
        "gpt-6-sol": {
          "name": "GPT-6 Sol",
          "limit": { "context": 1050000, "output": 128000 }
        },
        "claude-sonnet-latest": { "name": "Claude Sonnet (latest)" }
      }
    }
  }
}

OpenCode is built for many providers. Tool-calling quality can vary by model.

Compatibility#

Feature	Status	Notes
Streaming (SSE)	Yes	Chat Completions streaming is used as is.
Tool calling	Yes	OpenAI-format `tools` pass through. The model must support tool calls.
Image input	Yes	Works only when the selected model accepts images.
Automatic model list	Partial	OpenCode does not fetch a custom provider's list, so you list models in `models` yourself.

Desktop app and IDE extensions#

The OpenCode desktop app and IDE extensions read the same opencode.json and auth.json as the terminal. Once you are connected in the terminal, just pick a model — no extra setup. After editing the config, fully quit and reopen the app; the provider list is read only at startup.

Troubleshooting#

Symptom	Cause and fix
401 Unauthorized	`{env:PLEUM_API_KEY}` is empty or the key is wrong. Run `echo $PLEUM_API_KEY` in the same shell, and if you stored it with `/connect`, check the key in `auth.json`.
Not sure the key was saved	Run `opencode auth list` to see stored credentials. If `pleum` is missing, run `/connect` again.
A model you added is not in the list	Fully quit and reopen OpenCode (including the desktop app). Providers and models are read only at startup.
PleumRouter models are missing from `/models`	Check that you listed the IDs under `models` in `opencode.json` and that the provider ID (`pleum`) matches what you entered in `/connect`, then restart OpenCode.
I only ran `/connect` and nothing works	`/connect` stores only the key. `provider.pleum.options.baseURL` and `models` must be written in `opencode.json`.
404 Not Found	Check that `baseURL` is `https://apirouter.pleum.ai/v1`. Do not append `/chat/completions`.
model not found	An ID listed under `models` is not in the catalog. Confirm exact IDs with `GET /v1/models`.
Where should the config file go?	Use `./opencode.json` in the project root for one project, or `~/.config/opencode/opencode.json` for all projects.

Cost & security tips#

Set a budget on the key. A coding agent calls the model many times per task and resends long context. Daily, weekly or monthly budgets on your API key make costs easier to predict.
Keep the key out of git. Make sure a key in your shell profile or config file never gets committed, and if you suspect a leak, revoke it in the dashboard and issue a new one.
Check where sensitive code goes. Prompts and code are sent to the provider of the model you pick. Use Data residency and PII masking to control that.