OpenRouter Maestro

One extension. Three AI coding agents. 500+ models.
Run any OpenRouter model in GitHub Copilot Chat, Claude Code and OpenAI Codex — from a single panel.

Why

Every AI coding agent wants its own provider setup, and each one stores it somewhere different — a VS Code API, a JSON settings file, a TOML config. Maestro learns all three so you don't have to:

Everything it writes is reversible: a one-time backup, a snapshot of exactly the keys it touches, and a one-click switch back to the agent's own model.

Install

Search OpenRouter Maestro in the VS Code Extensions view, or:

code --install-extension husamettinulutas.openrouter-maestro

Quick start

1. Open the OpenRouter Maestro icon in the activity bar.
2. Click 🔑 and paste your OpenRouter API key — get one at openrouter.ai/settings/keys. It is stored in VS Code SecretStorage, never in a file.
3. Click 🔄 to sync the catalog, then browse: search, filter by vision / tools / free, sort by price or context length.
4. Every model card carries three buttons — ＋ Copilot, ＋ Claude Code, ＋ Codex. Press the ones you want. Each agent keeps its own saved list, so adding a model never drops the previous one.

Each agent then gets its own tab, and every tab shows the same card: capabilities, input/output price per million tokens, context window and max output.

Copilot Chat — every listed model is usable at once

They show up in the Copilot Chat model picker under OpenRouter Maestro immediately — no reload, no config file.

Claude Code — a saved list, one model live

Claude Code and Codex can each run one model at a time, so keep as many as you like in the list and press Activate on the one you want. The green card is the one that is really wired in.

Codex — and the way back

The dashed row at the bottom of every list — “…'s own model” — is the default. Pick it and Maestro removes its config, restores the original settings exactly as they were, and keeps your list for next time.

Reload the window after switching. Claude Code and Codex read their config when they start, so a session that is already open keeps the model it started with — including after you switch back to the default. Maestro shows a Reload Window button whenever you change something. A CLI running in a terminal has to be restarted on its own.

All of it is available from the command palette too: OpenRouter Maestro: … (Browse Models, Use Model in Claude Code, Use Model in Codex, Restore …, Show Integration Status).

How each integration works

// ~/.claude/settings.json  (or .claude/settings.local.json with the "project" scope setting)
{
  "env": {
    "ANTHROPIC_BASE_URL": "https://openrouter.ai/api",
    "ANTHROPIC_AUTH_TOKEN": "sk-or-v1-…",              // your OpenRouter key
    "ANTHROPIC_API_KEY": "",                            // explicitly empty — blocks fallback auth
    "ANTHROPIC_MODEL": "<openrouter-model-id>",
    "ANTHROPIC_DEFAULT_OPUS_MODEL": "<openrouter-model-id>",
    "ANTHROPIC_DEFAULT_SONNET_MODEL": "<openrouter-model-id>",
    "ANTHROPIC_DEFAULT_HAIKU_MODEL": "<small/fast model, or the same one>",
    "CLAUDE_CODE_SUBAGENT_MODEL": "<openrouter-model-id>",
    "CLAUDE_CODE_DISABLE_EXPERIMENTAL_BETAS": "1",      // beta request fields 400 on non-Anthropic gateways
    "CLAUDE_CODE_MAX_CONTEXT_TOKENS": "<model context>", // so "prompt is too long" can't beat compaction to it
    "CLAUDE_CODE_AUTO_COMPACT_WINDOW": "<~80% of context>",
    "CLAUDE_CODE_MAX_OUTPUT_TOKENS": "<min(model max, 32000)>"
  }
}

model = "<openrouter-model-id>"       # top-level keys must come before any [section]
model_provider = "openrouter"
show_raw_agent_reasoning = true       # written for reasoning-capable models
model_reasoning_effort = "high"       # minimal|low|medium|high|xhigh
model_reasoning_summary = "auto"

[model_providers.openrouter]
name = "OpenRouter"
base_url = "https://openrouter.ai/api/v1"
env_key = "OPENROUTER_API_KEY"
wire_api = "responses"

Good to know

• Uninstalling the extension does not undo the agent configs. VS Code runs no extension code on uninstall. Switch Claude Code and Codex back to their own model before uninstalling (or restore the .maestro-backup files by hand).
• Billing moves to OpenRouter. While an override is active, your Claude or ChatGPT subscription is not being used. Switching back to the default returns you to it.
• Your usage is labelled on the OpenRouter dashboard. Everything Maestro wires up — Copilot, Claude Code and Codex alike — sends HTTP-Referer and X-OpenRouter-Title, so Activity → Apps shows one OpenRouter Maestro entry that links back to this repo, instead of three piles of unattributed tokens. No usage data leaves your machine beyond the request itself.
• Rolling-alias ids start with ~ (e.g. ~deepseek/deepseek-v4-flash-latest). The tilde is part of the id and is written verbatim — the de-tilded form is rejected by OpenRouter.
• Codex hides thinking steps for most OpenRouter models, and no setting fixes it. Measured against Codex 0.148 / IDE 26.810: OpenRouter streams raw chain-of-thought as response.reasoning_text.delta / .done, while Codex's stream handler only recognises the summary events (response.reasoning_summary_text.delta, response.reasoning_summary_part.added) and drops the rest. The model does reason — turn.completed reports reasoning_output_tokens > 0, and you are billed for them — there is simply no event Codex will render. model_catalog_json (successor to the removed model_supports_reasoning_summaries key) was tested and changes nothing. Models whose provider emits real summaries display normally. This one needs a fix in Codex.
• Claude Code with non-Anthropic models: DeepSeek, Qwen, GLM and friends route fine, but tool-call translation can get imperfect in very long agentic loops.

Why not just Copilot's built-in OpenRouter BYOK?

VS Code's built-in BYOK has included OpenRouter since 2026, but it hides :free variant models, exposes no provider-routing or request options, and does nothing at all for Claude Code or Codex. Maestro shows the full catalog (free models included) with pricing and capability filters, adds prompt caching, guardrail-safe base64 sanitization and live cost readouts — and manages all three agents from one panel.

Settings

Development

npm install
npm run watch      # esbuild in watch mode — then press F5 for an Extension Development Host
npm run compile    # tsc --noEmit type check
npm run build      # production bundle
npm run package    # build a .vsix

Requires VS Code 1.104+ (the release where the language-model provider API was finalized).

Contributing

Issues and pull requests are welcome at github.com/husamettinulutas/openrouter-maestro. Bug reports are most useful with the agent involved, its version, and the relevant lines from the OpenRouter Maestro output channel.

Successor to openrouter-copilot-model-manager, extended from Copilot-only to all three agents.

License

MIT — see LICENSE.