Merge pull request #42 from MiniMax-AI/feat/M2.1

feat: M2.1

Author: AkairoDev

README.md

@@ -2,7 +2,7 @@
 
 English | [中文](./README_CN.md)
 
-**Mini Agent** is a minimal yet professional demo project that showcases the best practices for building agents with the MiniMax M2 model. Leveraging an Anthropic-compatible API, it fully supports interleaved thinking to unlock M2's powerful reasoning capabilities for long, complex tasks.
+**Mini Agent** is a minimal yet professional demo project that showcases the best practices for building agents with the MiniMax M2.1 model. Leveraging an Anthropic-compatible API, it fully supports interleaved thinking to unlock M2's powerful reasoning capabilities for long, complex tasks.
 
 This project comes packed with features designed for a robust and intelligent agent development experience:
 
@@ -115,7 +115,7 @@ Fill in your API Key and corresponding API Base:
 api_key: "YOUR_API_KEY_HERE"          # API Key from step 1
 api_base: "https://api.minimax.io"  # Global
 # api_base: "https://api.minimaxi.com"  # China
-model: "MiniMax-M2"
+model: "MiniMax-M2.1"
 ```
 
 **Start Using:**
@@ -182,7 +182,7 @@ Fill in your API Key and corresponding API Base:
 api_key: "YOUR_API_KEY_HERE"          # API Key from step 1
 api_base: "https://api.minimax.io"  # Global
 # api_base: "https://api.minimaxi.com"  # China
-model: "MiniMax-M2"
+model: "MiniMax-M2.1"
 max_steps: 100
 workspace_dir: "./workspace"
 ```

README_CN.md

@@ -2,7 +2,7 @@
 
 [English](./README.md) | 中文
 
-**Mini Agent** 是一个极简但专业的演示项目，旨在展示使用 MiniMax M2 模型构建 Agent 的最佳实践。项目通过兼容 Anthropic 的 API，完全支持交错思维（interleaved thinking），从而解锁 M2 模型在处理长而复杂的任务时强大的推理能力。
+**Mini Agent** 是一个极简但专业的演示项目，旨在展示使用 MiniMax M2.1 模型构建 Agent 的最佳实践。项目通过兼容 Anthropic 的 API，完全支持交错思维（interleaved thinking），从而解锁 M2 模型在处理长而复杂的任务时强大的推理能力。
 
 该项目具备一系列为稳健、智能的 Agent 开发而设计的特性：
 
@@ -115,7 +115,7 @@ nano ~/.mini-agent/config/config.yaml
 api_key: "YOUR_API_KEY_HERE"          # 填入第 1 步获取的 API Key
 api_base: "https://api.minimaxi.com"  # 国内版
 # api_base: "https://api.minimax.io"  # 海外版（如使用海外平台，请取消本行注释）
-model: "MiniMax-M2"
+model: "MiniMax-M2.1"
 ```
 
 **开始使用：**
@@ -182,7 +182,7 @@ vim mini_agent/config/config.yaml  # 或使用您偏好的编辑器
 api_key: "YOUR_API_KEY_HERE"          # 填入第 1 步获取的 API Key
 api_base: "https://api.minimaxi.com"  # 国内版
 # api_base: "https://api.minimax.io"  # 海外版（如使用海外平台，请修改此行）
-model: "MiniMax-M2"
+model: "MiniMax-M2.1"
 max_steps: 100
 workspace_dir: "./workspace"
 ```

docs/PRODUCTION_GUIDE.md

@@ -34,7 +34,7 @@ This project is a **teaching-level demo** that demonstrates the core concepts an
 
 ### 2.2 Model Fallback Mechanism
 
-Currently using a single fixed model (MiniMax-M2), which will directly report errors on failure.
+Currently using a single fixed model (MiniMax-M2.1), which will directly report errors on failure.
 
 - Introduce a model pool by configuring multiple model accounts to improve availability
 - Introduce automatic health checks, failure removal, circuit breaker strategies for the model pool

docs/PRODUCTION_GUIDE_CN.md

@@ -34,7 +34,7 @@
 
 ### 2.2 模型回退机制
 
-当前 Demo 固定使用单一模型（MiniMax-M2），调用失败时会直接报错。
+当前 Demo 固定使用单一模型（MiniMax-M2.1），调用失败时会直接报错。
 
 - **建立模型池**：配置多个模型账号，建立模型池以提高服务可用性。
 - **引入高可用策略**：为模型池引入自动健康检测、故障节点切换、熔断等高可用策略。

examples/05_provider_selection.py

@@ -28,7 +28,7 @@ async def demo_anthropic_provider():
     client = LLMClient(
         api_key=config["api_key"],
         provider=LLMProvider.ANTHROPIC,  # Specify Anthropic provider
-        model=config.get("model", "MiniMax-M2"),
+        model=config.get("model", "MiniMax-M2.1"),
     )
 
     print(f"Provider: {client.provider}")
@@ -63,7 +63,7 @@ async def demo_openai_provider():
     client = LLMClient(
         api_key=config["api_key"],
         provider=LLMProvider.OPENAI,  # Specify OpenAI provider
-        model=config.get("model", "MiniMax-M2"),
+        model=config.get("model", "MiniMax-M2.1"),
     )
 
     print(f"Provider: {client.provider}")
@@ -97,7 +97,7 @@ async def demo_default_provider():
     # Initialize client without specifying provider (defaults to Anthropic)
     client = LLMClient(
         api_key=config["api_key"],
-        model=config.get("model", "MiniMax-M2"),
+        model=config.get("model", "MiniMax-M2.1"),
     )
 
     print(f"Provider (default): {client.provider}")
@@ -130,13 +130,13 @@ async def demo_provider_comparison():
     anthropic_client = LLMClient(
         api_key=config["api_key"],
         provider=LLMProvider.ANTHROPIC,
-        model=config.get("model", "MiniMax-M2"),
+        model=config.get("model", "MiniMax-M2.1"),
     )
 
     openai_client = LLMClient(
         api_key=config["api_key"],
         provider=LLMProvider.OPENAI,
-        model=config.get("model", "MiniMax-M2"),
+        model=config.get("model", "MiniMax-M2.1"),
     )
 
     # Same question for both

examples/06_tool_schema_demo.py

@@ -166,7 +166,7 @@ async def demo_tool_schemas():
     client = LLMClient(
         api_key=config["api_key"],
         provider=LLMProvider.ANTHROPIC,
-        model="MiniMax-M2",
+        model="MiniMax-M2.1",
     )
 
     # Test with a query that should trigger weather tool
@@ -215,7 +215,7 @@ async def demo_multiple_tools():
     client = LLMClient(
         api_key=config["api_key"],
         provider=LLMProvider.ANTHROPIC,
-        model="MiniMax-M2",
+        model="MiniMax-M2.1",
     )
 
     messages = [Message(role="user", content="Calculate 15 * 23 for me")]

mini_agent/cli.py

@@ -418,7 +418,7 @@ def on_retry(exception: Exception, attempt: int):
         system_prompt = system_prompt_path.read_text(encoding="utf-8")
         print(f"{Colors.GREEN}✅ Loaded system prompt (from: {system_prompt_path}){Colors.RESET}")
     else:
-        system_prompt = "You are Mini-Agent, an intelligent assistant powered by MiniMax M2 that can help users complete various tasks."
+        system_prompt = "You are Mini-Agent, an intelligent assistant powered by MiniMax M2.1 that can help users complete various tasks."
         print(f"{Colors.YELLOW}⚠️  System prompt not found, using default{Colors.RESET}")
 
     # 6. Inject Skills Metadata into System Prompt (Progressive Disclosure - Level 1)

mini_agent/config.py

@@ -24,7 +24,7 @@ class LLMConfig(BaseModel):
 
     api_key: str
     api_base: str = "https://api.minimax.io"
-    model: str = "MiniMax-M2"
+    model: str = "MiniMax-M2.1"
     provider: str = "anthropic"  # "anthropic" or "openai"
     retry: RetryConfig = Field(default_factory=RetryConfig)
 
@@ -116,7 +116,7 @@ def from_yaml(cls, config_path: str | Path) -> "Config":
         llm_config = LLMConfig(
             api_key=data["api_key"],
             api_base=data.get("api_base", "https://api.minimax.io"),
-            model=data.get("model", "MiniMax-M2"),
+            model=data.get("model", "MiniMax-M2.1"),
             provider=data.get("provider", "anthropic"),
             retry=retry_config,
         )

mini_agent/config/config-example.yaml

@@ -19,7 +19,7 @@
 api_key: "YOUR_API_KEY_HERE"  # Replace with your MiniMax API Key
 api_base: "https://api.minimax.io"   # Global users (default)
 # api_base: "https://api.minimaxi.com"  # China users
-model: "MiniMax-M2"
+model: "MiniMax-M2.1"
 # LLM provider: "anthropic" or "openai"
 # The LLMClient will automatically append /anthropic or /v1 to api_base based on provider
 provider: "anthropic"  # Default: anthropic

mini_agent/llm/anthropic_client.py

@@ -25,15 +25,15 @@ def __init__(
         self,
         api_key: str,
         api_base: str = "https://api.minimaxi.com/anthropic",
-        model: str = "MiniMax-M2",
+        model: str = "MiniMax-M2.1",
         retry_config: RetryConfig | None = None,
     ):
         """Initialize Anthropic client.
 
         Args:
             api_key: API key for authentication
             api_base: Base URL for the API (default: MiniMax Anthropic endpoint)
-            model: Model name to use (default: MiniMax-M2)
+            model: Model name to use (default: MiniMax-M2.1)
             retry_config: Optional retry configuration
         """
         super().__init__(api_key, api_base, model, retry_config)