opencode-copilot-1m

OpenCode plugin that adds long-context variants for Claude and GPT-5.4 models on GitHub Copilot.

Without this plugin, Copilot can expose smaller client-side limits than you want for long sessions. This plugin creates separate (1M) model entries with higher client-side limits while keeping the original models available.

Supported models

• Claude Opus 4.6
• Claude Opus 4.7
• Claude Sonnet 4.6
• GPT-5.4

Install

Add to your opencode.json:

{
  "plugin": ["opencode-copilot-1m"]
}

Restart OpenCode. You'll see new model entries like "Claude Opus 4.7 (1M)" and "GPT-5.4 (1M)" alongside the originals.

Recommended: set reasoning effort

The plugin can't control reasoning effort (that's a per-user config setting). Recommended settings for the 1M variants:

{
  "plugin": ["opencode-copilot-1m"],
  "provider": {
    "github-copilot": {
      "models": {
        "claude-opus-4.6-1m": {
          "reasoningEffort": "high"
        },
        "claude-opus-4.7-1m": {
          "reasoningEffort": "medium"
        }
      }
    }
  }
}

high is the max for Opus 4.6. medium is currently the highest available for Opus 4.7.

How it works

The plugin hooks into the github-copilot provider's model list and clones each supported model into a -1m variant with model-specific long-context limits. It keeps the original API model ID but overrides the client-side limit metadata OpenCode uses for budgeting and compaction.

It also sets the Copilot-Integration-Id: copilot-developer-cli header on all Copilot requests.

Related

• opencode-anthropic-context-1m -- same thing for Anthropic direct API and Amazon Bedrock
• OpenCode issue #12338 -- background on the problem
• OpenCode issue #16298 -- GPT-5.4 input/context limit background

License

MIT