feat: add MiniMax as first-class LLM provider

README.md

@@ -206,6 +206,7 @@ We support all major LLM providers and models for inference. You need to set the
 |----------|-------------------------------|------------------|
 | OptiLLM | `OPTILLM_API_KEY` | Uses the inbuilt local server for inference, supports logprobs and decoding techniques like `cot_decoding` & `entropy_decoding` |
 | OpenAI | `OPENAI_API_KEY` | You can use this with any OpenAI compatible endpoint (e.g. OpenRouter) by setting the `base_url` |
+| MiniMax | `MINIMAX_API_KEY` | Uses the [MiniMax API](https://www.minimax.io/) (OpenAI-compatible). Supports MiniMax-M2.7 and other models. Temperature is auto-clamped to (0, 1] |
 | Cerebras | `CEREBRAS_API_KEY` | You can use this for fast inference with supported models, see [docs for details](https://inference-docs.cerebras.ai/introduction) |
 | Azure OpenAI | `AZURE_OPENAI_API_KEY`<br>`AZURE_API_VERSION`<br>`AZURE_API_BASE` | - |
 | Azure OpenAI (Managed Identity) | `AZURE_API_VERSION`<br>`AZURE_API_BASE` | Login required using `az login`, see [docs for details](https://learn.microsoft.com/en-us/azure/ai-services/openai/how-to/managed-identity) |

optillm/server.py

@@ -91,6 +91,13 @@ def get_config():
             default_client = Cerebras(api_key=API_KEY, base_url=base_url, http_client=http_client)
         else:
             default_client = Cerebras(api_key=API_KEY, http_client=http_client)
+    elif os.environ.get("MINIMAX_API_KEY"):
+        API_KEY = os.environ.get("MINIMAX_API_KEY")
+        base_url = server_config['base_url']
+        if base_url == "":
+            base_url = "https://api.minimax.io/v1"
+        default_client = OpenAI(api_key=API_KEY, base_url=base_url, http_client=http_client)
+        logger.info(f"Created MiniMax client with base_url: {base_url}")
     elif os.environ.get("OPENAI_API_KEY"):
         API_KEY = os.environ.get("OPENAI_API_KEY")
         base_url = server_config['base_url']
@@ -759,6 +766,15 @@ def proxy():
     base_url = server_config['base_url']
     default_client, api_key = get_config()
 
+    # Clamp temperature for MiniMax provider: must be in (0.0, 1.0]
+    if os.environ.get("MINIMAX_API_KEY") and 'temperature' in request_config:
+        temp = request_config['temperature']
+        if temp is not None:
+            if temp <= 0:
+                request_config['temperature'] = 0.01
+            elif temp > 1.0:
+                request_config['temperature'] = 1.0
+
     operation, approaches, model = parse_combined_approach(model, known_approaches, plugin_approaches)
 
     # Start conversation logging if enabled

tests/test_minimax_integration.py

@@ -0,0 +1,87 @@
+"""
+Integration tests for MiniMax provider support in optillm.
+
+These tests verify that the MiniMax provider works end-to-end with the actual
+MiniMax API. They require a valid MINIMAX_API_KEY environment variable.
+
+Run with: MINIMAX_API_KEY=your-key pytest tests/test_minimax_integration.py -v
+"""
+
+import unittest
+import os
+import sys
+
+# Add parent directory to path to import optillm modules
+sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+
+# Skip all tests if MINIMAX_API_KEY is not set
+MINIMAX_API_KEY = os.environ.get('MINIMAX_API_KEY', '')
+SKIP_REASON = "MINIMAX_API_KEY environment variable not set"
+
+
+@unittest.skipUnless(MINIMAX_API_KEY, SKIP_REASON)
+class TestMiniMaxIntegration(unittest.TestCase):
+    """Integration tests that call the actual MiniMax API."""
+
+    def setUp(self):
+        """Set up OpenAI client pointing to MiniMax API."""
+        from openai import OpenAI
+        self.client = OpenAI(
+            api_key=MINIMAX_API_KEY,
+            base_url="https://api.minimax.io/v1"
+        )
+
+    def test_basic_completion(self):
+        """Test basic chat completion with MiniMax API."""
+        response = self.client.chat.completions.create(
+            model="MiniMax-M2.7",
+            messages=[
+                {"role": "user", "content": "Say hello in one word."}
+            ],
+            max_tokens=10,
+            temperature=0.7
+        )
+
+        assert hasattr(response, 'choices')
+        assert len(response.choices) > 0
+        assert response.choices[0].message.content is not None
+        assert len(response.choices[0].message.content) > 0
+
+    def test_temperature_boundary(self):
+        """Test that MiniMax API accepts temperature at boundary value 0.01."""
+        response = self.client.chat.completions.create(
+            model="MiniMax-M2.7",
+            messages=[
+                {"role": "user", "content": "What is 2+2?"}
+            ],
+            max_tokens=10,
+            temperature=0.01
+        )
+
+        assert hasattr(response, 'choices')
+        assert len(response.choices) > 0
+
+    def test_streaming_completion(self):
+        """Test streaming chat completion with MiniMax API."""
+        stream = self.client.chat.completions.create(
+            model="MiniMax-M2.7",
+            messages=[
+                {"role": "user", "content": "Count from 1 to 3."}
+            ],
+            max_tokens=30,
+            temperature=0.5,
+            stream=True
+        )
+
+        chunks = list(stream)
+        assert len(chunks) > 0
+        content_chunks = [
+            chunk.choices[0].delta.content
+            for chunk in chunks
+            if chunk.choices[0].delta.content
+        ]
+        assert len(content_chunks) > 0
+
+
+if __name__ == '__main__':
+    unittest.main()