Add MiniMax as LLM provider via OpenAI-compatible API

docs/integrations/llms/index.md

@@ -6,4 +6,5 @@ This section provides tutorials on incorporating alternative large language mode
 :maxdepth: 1
 :hidden:
 
+minimax
 ```

docs/integrations/llms/minimax.md

@@ -0,0 +1,106 @@
+# MiniMax
+
+[MiniMax](https://www.minimaxi.com/) provides large language models accessible through an OpenAI-compatible API. This guide shows how to integrate MiniMax models into your prompt flow.
+
+## Prerequisites
+
+- A MiniMax API key from [MiniMax Platform](https://platform.minimaxi.com/)
+
+## Available Models
+
+| Model | Context Window | Description |
+|-------|---------------|-------------|
+| `MiniMax-M2.5` | 204K tokens | General-purpose model |
+| `MiniMax-M2.5-highspeed` | 204K tokens | Faster variant for lower latency |
+
+## Setup Connection
+
+MiniMax uses an OpenAI-compatible API, so you can use prompt flow's built-in `OpenAIConnection` with a custom `base_url`.
+
+### Using CLI
+
+```bash
+pf connection create -f minimax.yml --set api_key=<your-minimax-api-key>
+```
+
+Where `minimax.yml` contains:
+
+```yaml
+$schema: https://azuremlschemas.azureedge.net/promptflow/latest/OpenAIConnection.schema.json
+name: minimax_connection
+type: open_ai
+api_key: "<user-input>"
+base_url: "https://api.minimax.io/v1"
+```
+
+### Using Python SDK
+
+```python
+from promptflow.core import OpenAIModelConfiguration
+
+config = OpenAIModelConfiguration(
+    model="MiniMax-M2.5",
+    base_url="https://api.minimax.io/v1",
+    api_key="your-minimax-api-key",
+)
+```
+
+### Using Environment Variables
+
+```bash
+export MINIMAX_API_KEY="your-api-key"
+```
+
+```python
+import os
+from promptflow.core import OpenAIModelConfiguration
+
+config = OpenAIModelConfiguration(
+    model="MiniMax-M2.5",
+    base_url="https://api.minimax.io/v1",
+    api_key=os.environ["MINIMAX_API_KEY"],
+)
+```
+
+## Use in a Prompty File
+
+```yaml
+---
+name: MiniMax Chat
+model:
+  api: chat
+  configuration:
+    type: openai
+    model: MiniMax-M2.5
+    base_url: https://api.minimax.io/v1
+  parameters:
+    temperature: 0.7
+    max_tokens: 1024
+---
+```
+
+## Use in a Flex Flow
+
+```python
+from promptflow.core import OpenAIModelConfiguration, Prompty
+from promptflow.tracing import trace
+
+config = OpenAIModelConfiguration(
+    model="MiniMax-M2.5",
+    base_url="https://api.minimax.io/v1",
+    api_key="your-api-key",
+)
+
+prompty = Prompty.load(source="chat.prompty", model={"configuration": config})
+result = prompty(question="Hello!")
+```
+
+## Notes
+
+- **Temperature**: MiniMax accepts temperature values in the range `[0.0, 1.0]`.
+- **Context window**: Both MiniMax-M2.5 models support up to 204K tokens.
+- **Streaming**: Supported via the standard OpenAI streaming interface.
+
+## Example
+
+See the complete working example at [examples/flex-flows/chat-with-minimax](../../examples/flex-flows/chat-with-minimax/).

examples/connections/minimax.yml

@@ -0,0 +1,5 @@
+$schema: https://azuremlschemas.azureedge.net/promptflow/latest/OpenAIConnection.schema.json
+name: minimax_connection
+type: open_ai
+api_key: "<user-input>"  # Your MiniMax API key from https://platform.minimaxi.com
+base_url: "https://api.minimax.io/v1"

examples/flex-flows/chat-with-minimax/README.md

@@ -0,0 +1,58 @@
+# Chat with MiniMax
+
+This example demonstrates how to use [MiniMax](https://www.minimaxi.com/) as an LLM provider in prompt flow. MiniMax provides an OpenAI-compatible API, so it integrates seamlessly with prompt flow's existing OpenAI connection type.
+
+## Prerequisites
+
+- Install prompt flow: `pip install promptflow promptflow-tools`
+- A MiniMax API key from [MiniMax Platform](https://platform.minimaxi.com/)
+
+## Available Models
+
+| Model | Description |
+|-------|-------------|
+| `MiniMax-M2.5` | General-purpose model with 204K context window |
+| `MiniMax-M2.5-highspeed` | Faster variant optimized for lower latency |
+
+## Setup
+
+### Option 1: Using environment variable
+
+```bash
+export MINIMAX_API_KEY="your-api-key-here"
+```
+
+### Option 2: Using a prompt flow connection
+
+Create a MiniMax connection using the OpenAI connection type with a custom base URL:
+
+```bash
+pf connection create -f ../../connections/minimax.yml --set api_key=<your-api-key>
+```
+
+## Run the example
+
+### Run directly
+
+```bash
+export MINIMAX_API_KEY="your-api-key-here"
+python flow.py
+```
+
+### Run as a flow
+
+```bash
+pf flow test --flow . --inputs question="What is Prompt flow?"
+```
+
+### Run with batch data
+
+```bash
+pf run create --flow . --data data.jsonl --column-mapping question='${data.question}' --stream
+```
+
+## Notes
+
+- MiniMax's API is OpenAI-compatible, so it works with prompt flow's `OpenAIConnection` by setting `base_url` to `https://api.minimax.io/v1`.
+- Temperature values are accepted in the range `[0.0, 1.0]`.
+- The `MiniMax-M2.5` model supports a 204K context window, suitable for long-document analysis.

examples/flex-flows/chat-with-minimax/chat.prompty

@@ -0,0 +1,31 @@
+---
+name: MiniMax Chat
+model:
+  api: chat
+  configuration:
+    type: openai
+    model: MiniMax-M2.5
+    base_url: https://api.minimax.io/v1
+  parameters:
+    temperature: 0.7
+    max_tokens: 1024
+inputs:
+  question:
+    type: string
+  chat_history:
+    type: list
+sample:
+  question: "What is Prompt flow?"
+  chat_history: []
+---
+
+system:
+You are a helpful assistant.
+
+{% for item in chat_history %}
+{{item.role}}:
+{{item.content}}
+{% endfor %}
+
+user:
+{{question}}

examples/flex-flows/chat-with-minimax/data.jsonl

@@ -0,0 +1,3 @@
+{"question": "What is Prompt flow?"}
+{"question": "How do I create a flow?"}
+{"question": "What are the benefits of using Prompt flow?"}

examples/flex-flows/chat-with-minimax/flow.flex.yaml

@@ -0,0 +1,12 @@
+$schema: https://azuremlschemas.azureedge.net/promptflow/latest/Flow.schema.json
+entry: flow:ChatFlow
+sample:
+  inputs:
+    question: What is Prompt flow?
+  init:
+    model_config:
+      connection: minimax_connection
+      model: MiniMax-M2.5
+    max_total_token: 4096
+environment:
+  python_requirements_txt: requirements.txt

examples/flex-flows/chat-with-minimax/flow.py

@@ -0,0 +1,98 @@
+import os
+from pathlib import Path
+
+from promptflow.tracing import trace
+from promptflow.core import OpenAIModelConfiguration, Prompty
+
+BASE_DIR = Path(__file__).absolute().parent
+
+# MiniMax API base URL (OpenAI-compatible)
+MINIMAX_BASE_URL = "https://api.minimax.io/v1"
+
+# Available MiniMax models
+MINIMAX_MODELS = {
+    "MiniMax-M2.5": "General-purpose model with 204K context window",
+    "MiniMax-M2.5-highspeed": "Faster variant optimized for lower latency",
+}
+
+
+def _clamp_temperature(temperature: float) -> float:
+    """Clamp temperature to MiniMax's accepted range [0.0, 1.0]."""
+    return max(0.0, min(1.0, temperature))
+
+
+class ChatFlow:
+    """A chat flow powered by MiniMax's LLM via the OpenAI-compatible API.
+
+    MiniMax provides an OpenAI-compatible API endpoint, so it works seamlessly
+    with prompt flow's OpenAI connection type by setting the base_url.
+    """
+
+    def __init__(
+        self,
+        model_config: OpenAIModelConfiguration,
+        max_total_token: int = 4096,
+    ):
+        self.model_config = model_config
+        self.max_total_token = max_total_token
+
+    @trace
+    def __call__(
+        self,
+        question: str = "What is Prompt flow?",
+        chat_history: list = None,
+    ) -> str:
+        """Flow entry function."""
+        prompty = Prompty.load(
+            source=BASE_DIR / "chat.prompty",
+            model={"configuration": self.model_config},
+        )
+
+        chat_history = chat_history or []
+        while len(chat_history) > 0:
+            token_count = prompty.estimate_token_count(
+                question=question, chat_history=chat_history
+            )
+            if token_count > self.max_total_token:
+                chat_history = chat_history[1:]
+            else:
+                break
+
+        output = prompty(question=question, chat_history=chat_history)
+        return output
+
+
+def get_minimax_config(
+    model: str = "MiniMax-M2.5",
+    api_key: str = None,
+) -> OpenAIModelConfiguration:
+    """Create an OpenAIModelConfiguration pre-configured for MiniMax.
+
+    Args:
+        model: MiniMax model name. Options: MiniMax-M2.5, MiniMax-M2.5-highspeed.
+        api_key: MiniMax API key. Falls back to MINIMAX_API_KEY env var.
+
+    Returns:
+        OpenAIModelConfiguration configured for MiniMax.
+    """
+    api_key = api_key or os.environ.get("MINIMAX_API_KEY")
+    if not api_key:
+        raise ValueError(
+            "MiniMax API key is required. Set MINIMAX_API_KEY environment variable "
+            "or pass api_key parameter."
+        )
+    return OpenAIModelConfiguration(
+        model=model,
+        base_url=MINIMAX_BASE_URL,
+        api_key=api_key,
+    )
+
+
+if __name__ == "__main__":
+    from promptflow.tracing import start_trace
+
+    start_trace()
+    config = get_minimax_config(model="MiniMax-M2.5")
+    flow = ChatFlow(config)
+    result = flow("What is Prompt flow?", [])
+    print(result)

examples/flex-flows/chat-with-minimax/requirements.txt

@@ -0,0 +1,4 @@
+promptflow
+promptflow-tools
+openai>=1.0.0
+python-dotenv