feat: add OrcaRouter as a named backend

Add `orcarouter` as a first-class backend mirroring the existing
`openrouter` wiring: it dispatches to OpenAIClient with a default
base_url of https://api.orcarouter.ai/v1 and auto-loads
ORCAROUTER_API_KEY when no api_key is passed.

Also updates the ClientBackend literal, README, and the docs
(backends, api reference, and getting-started).

Author: zsanig22-dotcom

README.md

@@ -127,7 +127,7 @@ export PRIME_API_KEY=...
 
 
 ### Model Providers
-We currently support most major clients (OpenAI, Anthropic), as well as the router platforms (OpenRouter, Portkey). For local models, we recommend using vLLM (which interfaces with the [OpenAI client](https://github.com/alexzhang13/rlm/blob/main/rlm/clients/openai.py)). To view or add support for more clients, start by looking at [`rlm/clients/`](https://github.com/alexzhang13/rlm/tree/main/rlm/clients).
+We currently support most major clients (OpenAI, Anthropic), as well as the router platforms (OpenRouter, Portkey, OrcaRouter). For local models, we recommend using vLLM (which interfaces with the [OpenAI client](https://github.com/alexzhang13/rlm/blob/main/rlm/clients/openai.py)). To view or add support for more clients, start by looking at [`rlm/clients/`](https://github.com/alexzhang13/rlm/tree/main/rlm/clients).
 
 ## Training
 We provide a simple RL training harness for training RLMs used in this repo (specifically the `local` REPL). The implementation uses no sandboxes for simplicity and slots easily your use case, but an ideal setup would use sandboxes for safety. Training logic is isolated to the [`training/`](https://github.com/alexzhang13/rlm/tree/main/training) folder, which exposes `rlm.RLM` as a [`verifiers`](https://github.com/willccbb/verifiers) `Environment` and plugs straight into [`prime-rl`](https://github.com/PrimeIntellect-ai/prime-rl). See the [training README](https://github.com/alexzhang13/rlm/tree/main/training#readme) for the launch command. The harness uses subprocess-isolated local REPL execution (no cloud sandboxes), matching the `local` environment above.

docs/api/rlm.md

@@ -71,7 +71,7 @@ RLM(
 #### `backend`
 {: .no_toc }
 
-**Type:** `Literal["openai", "portkey", "openrouter", "vllm", "anthropic"]`
+**Type:** `Literal["openai", "portkey", "openrouter", "orcarouter", "vllm", "anthropic"]`
 **Default:** `"openai"`
 
 The LM provider backend to use for the root model.
@@ -103,6 +103,7 @@ Configuration passed to the LM client. Required fields vary by backend:
 | `anthropic` | `model_name` | `api_key` |
 | `portkey` | `model_name`, `api_key` | `base_url` |
 | `openrouter` | `model_name` | `api_key` |
+| `orcarouter` | `model_name` | `api_key` |
 | `vllm` | `model_name`, `base_url` | — |
 
 

docs/getting-started.md

@@ -317,6 +317,18 @@ rlm = RLM(
 )
 ```
 
+### OrcaRouter
+
+```python
+rlm = RLM(
+    backend="orcarouter",
+    backend_kwargs={
+        "api_key": os.getenv("ORCAROUTER_API_KEY"),
+        "model_name": "orcarouter/auto",  # Adaptive routing, or use vendor/model
+    },
+)
+```
+
 ### vLLM (Local)
 
 ```python

docs/src/app/api/page.tsx

@@ -61,6 +61,7 @@ print(result.response)`} />
                 [<code key="2" className="text-sm">"anthropic"</code>, "Anthropic API"],
                 [<code key="3" className="text-sm">"portkey"</code>, "Portkey AI gateway"],
                 [<code key="4" className="text-sm">"openrouter"</code>, "OpenRouter"],
+                [<code key="5" className="text-sm">"orcarouter"</code>, "OrcaRouter"],
 
                 [<code key="6" className="text-sm">"vllm"</code>, "Local vLLM server"],
               ]}

docs/src/app/backends/page.tsx

@@ -64,6 +64,20 @@ export default function BackendsPage() {
 
       <hr className="my-8 border-border" />
 
+      <h2 className="text-2xl font-semibold mb-4">OrcaRouter</h2>
+      <p className="text-muted-foreground mb-4">
+        <a href="https://www.orcarouter.ai" className="text-primary underline font-medium" target="_blank" rel="noopener noreferrer">OrcaRouter</a> is an OpenAI-compatible AI gateway that routes requests at upstream cost across 150+ models from OpenAI, Anthropic, Google Gemini, DeepSeek, and more — with adaptive routing, automatic failover, and observability.
+      </p>
+      <CodeBlock code={`rlm = RLM(
+    backend="orcarouter",
+    backend_kwargs={
+        "api_key": os.getenv("ORCAROUTER_API_KEY"),
+        "model_name": "anthropic/claude-opus-4.8",  # Format: vendor/model, or orcarouter/auto
+    },
+)`} />
+
+      <hr className="my-8 border-border" />
+
       <h2 className="text-2xl font-semibold mb-4">vLLM (Local)</h2>
       <p className="text-muted-foreground mb-4">Local model serving.</p>
       <CodeBlock language="bash" code={`# Start vLLM server

rlm/clients/__init__.py

@@ -36,6 +36,11 @@ def get_client(
 
         backend_kwargs.setdefault("base_url", "https://openrouter.ai/api/v1")
         return OpenAIClient(**backend_kwargs)
+    elif backend == "orcarouter":
+        from rlm.clients.openai import OpenAIClient
+
+        backend_kwargs.setdefault("base_url", "https://api.orcarouter.ai/v1")
+        return OpenAIClient(**backend_kwargs)
     elif backend == "vercel":
         from rlm.clients.openai import OpenAIClient
 
@@ -55,5 +60,5 @@ def get_client(
         return AzureOpenAIClient(**backend_kwargs)
     else:
         raise ValueError(
-            f"Unknown backend: {backend}. Supported backends: ['openai', 'vllm', 'portkey', 'openrouter', 'anthropic', 'azure_openai', 'gemini', 'vercel']"
+            f"Unknown backend: {backend}. Supported backends: ['openai', 'vllm', 'portkey', 'openrouter', 'orcarouter', 'anthropic', 'azure_openai', 'gemini', 'vercel']"
         )

rlm/clients/openai.py

@@ -15,6 +15,7 @@
 DEFAULT_OPENROUTER_API_KEY = os.getenv("OPENROUTER_API_KEY")
 DEFAULT_VERCEL_API_KEY = os.getenv("AI_GATEWAY_API_KEY")
 DEFAULT_PRIME_API_KEY = os.getenv("PRIME_API_KEY")
+DEFAULT_ORCAROUTER_API_KEY = os.getenv("ORCAROUTER_API_KEY")
 DEFAULT_PRIME_INTELLECT_BASE_URL = "https://api.pinference.ai/api/v1/"
 
 
@@ -66,6 +67,8 @@ def __init__(
                 api_key = DEFAULT_OPENAI_API_KEY
             elif base_url == "https://openrouter.ai/api/v1":
                 api_key = DEFAULT_OPENROUTER_API_KEY
+            elif base_url == "https://api.orcarouter.ai/v1":
+                api_key = DEFAULT_ORCAROUTER_API_KEY
             elif base_url == "https://ai-gateway.vercel.sh/v1":
                 api_key = DEFAULT_VERCEL_API_KEY
             elif base_url == DEFAULT_PRIME_INTELLECT_BASE_URL:

rlm/core/types.py

@@ -6,6 +6,7 @@
     "openai",
     "portkey",
     "openrouter",
+    "orcarouter",
     "vercel",
     "vllm",
     "anthropic",

tests/test_imports.py

@@ -73,6 +73,19 @@ def test_get_client_function(self):
 
         assert callable(get_client)
 
+    def test_orcarouter_backend_dispatch(self):
+        """Test that get_client routes the orcarouter backend to OpenAIClient."""
+        pytest.importorskip("openai")
+        from rlm.clients import get_client
+        from rlm.clients.openai import OpenAIClient
+
+        client = get_client(
+            "orcarouter",
+            {"api_key": "sk-test", "model_name": "orcarouter/auto"},
+        )
+        assert isinstance(client, OpenAIClient)
+        assert client.base_url == "https://api.orcarouter.ai/v1"
+
 
 class TestCoreImports:
     """Test core module imports."""