UnicomAI · octo-patch · Mar 22, 2026
diff --git a/README.md b/README.md
@@ -63,7 +63,7 @@
 ### 🚩 Core Function Modules
 
 #### **1. Model Management (Model Hub)**
-▸ Supports the unified access and lifecycle management of **hundreds of proprietary/open-source large models** (including GPT, Claude, Llama, etc.)
+▸ Supports the unified access and lifecycle management of **hundreds of proprietary/open-source large models** (including GPT, Claude, Llama, MiniMax, etc.)
 
 ▸ Deeply adapts to **OpenAI API standards** and **Unicom Yuanjing** ecological models, realizing seamless switching of heterogeneous models
 
@@ -285,7 +285,7 @@ To help you quickly get started with this project, we strongly recommend that yo
 
 | Feature                                                      | Detailed Description                                         |
 | :----------------------------------------------------------- | :----------------------------------------------------------- |
-| [Model Management](https://github.com/UnicomAI/wanwu/blob/main/configs/microservice/bff-service/static/manual/1.%E6%A8%A1%E5%9E%8B%E7%AE%A1%E7%90%86.md) | Supports users to import LLM, Embedding, and Rerank models from various model providers, including Unicom Yuanjing, OpenAI-API-compatible, Ollama, Tongyi Qianwen, and Volcano Engine. [Model Import Methods - Detailed Version](https://github.com/UnicomAI/wanwu/blob/main/configs/microservice/bff-service/static/manual/%E6%A8%A1%E5%9E%8B%E5%AF%BC%E5%85%A5%E6%96%B9%E5%BC%8F-%E8%AF%A6%E7%BB%86%E7%89%88.md) |
+| [Model Management](https://github.com/UnicomAI/wanwu/blob/main/configs/microservice/bff-service/static/manual/1.%E6%A8%A1%E5%9E%8B%E7%AE%A1%E7%90%86.md) | Supports users to import LLM, Embedding, and Rerank models from various model providers, including Unicom Yuanjing, OpenAI-API-compatible, Ollama, Tongyi Qianwen, Volcano Engine, and MiniMax. [Model Import Methods - Detailed Version](https://github.com/UnicomAI/wanwu/blob/main/configs/microservice/bff-service/static/manual/%E6%A8%A1%E5%9E%8B%E5%AF%BC%E5%85%A5%E6%96%B9%E5%BC%8F-%E8%AF%A6%E7%BB%86%E7%89%88.md) |
 | [Knowledge Base](https://github.com/UnicomAI/wanwu/tree/main/configs/microservice/bff-service/static/manual/2.%E7%9F%A5%E8%AF%86%E5%BA%93) | In terms of document parsing capabilities: supports uploading of 12 file types and URL parsing; Supports private deployment and integration for document parsing via two methods: OCR and [a proprietary MinerU model (for scenarios like titles, tables, and formulas)](https://github.com/UnicomAI/DocParserServer/tree/main) ; document segmentation settings support both general segmentation and parent-child segmentation. In terms of optimization capabilities: supports metadata management 、Graph RAG and metadata filtering queries, supports adding, deleting, and modifying segmented content, supports setting keyword tags for segments to improve recall performance, supports segment enable/disable operations, and supports hit testing. In terms of retrieval capabilities: supports multiple retrieval modes including vector search, full-text search, and hybrid search. In terms of Q&A capabilities: supports automatic citation of sources and generating answers with both text and images.<br |
 | [Resource Library](https://github.com/UnicomAI/wanwu/blob/main/configs/microservice/bff-service/static/manual/3.%E5%B7%A5%E5%85%B7%E5%B9%BF%E5%9C%BA.md) | Supports importing your own MCP services or custom tools for use in workflows and agents. |
 | [Safety Guardrails](https://github.com/UnicomAI/wanwu/blob/main/configs/microservice/bff-service/static/manual/4.%E5%AE%89%E5%85%A8%E6%8A%A4%E6%A0%8F.md) | Users can create sensitive word lists to control the safety of the model's output. |

diff --git a/README_CN.md b/README_CN.md
@@ -66,7 +66,7 @@
 
 #### **1. 模型纳管（Model Hub）**
 
-▸ 支持 **数百种专有/开源大模型**（包括GPT、Claude、Llama等系列）的统一接入与生命周期管理
+▸ 支持 **数百种专有/开源大模型**（包括GPT、Claude、Llama、MiniMax等系列）的统一接入与生命周期管理
 
 ▸ 深度适配 **OpenAI API 标准** 及 **联通元景** 生态模型，实现异构模型的无缝切换
 
@@ -303,7 +303,7 @@
 
 |                             功能                             |                           详细描述                           |
 | :----------------------------------------------------------: | :----------------------------------------------------------: |
-| [模型管理](https://github.com/UnicomAI/wanwu/blob/main/configs/microservice/bff-service/static/manual/1.%E6%A8%A1%E5%9E%8B%E7%AE%A1%E7%90%86.md) | 支持用户导入包括联通元景、OpenAI-API-compatible、Ollama、通义千问、火山引擎等模型供应商的LLM、Embedding、Rerank模型。[ 模型导入方式-详细版](https://github.com/UnicomAI/wanwu/blob/main/configs/microservice/bff-service/static/manual/%E6%A8%A1%E5%9E%8B%E5%AF%BC%E5%85%A5%E6%96%B9%E5%BC%8F-%E8%AF%A6%E7%BB%86%E7%89%88.md) |
+| [模型管理](https://github.com/UnicomAI/wanwu/blob/main/configs/microservice/bff-service/static/manual/1.%E6%A8%A1%E5%9E%8B%E7%AE%A1%E7%90%86.md) | 支持用户导入包括联通元景、OpenAI-API-compatible、Ollama、通义千问、火山引擎、MiniMax等模型供应商的LLM、Embedding、Rerank模型。[ 模型导入方式-详细版](https://github.com/UnicomAI/wanwu/blob/main/configs/microservice/bff-service/static/manual/%E6%A8%A1%E5%9E%8B%E5%AF%BC%E5%85%A5%E6%96%B9%E5%BC%8F-%E8%AF%A6%E7%BB%86%E7%89%88.md) |
 | [知识库](https://github.com/UnicomAI/wanwu/tree/main/configs/microservice/bff-service/static/manual/2.%E7%9F%A5%E8%AF%86%E5%BA%93) | 在文档解析能力方面:支持12种文件类型的上传，支持url解析;文档解析方式支持OCR和[ MinerU模型解析(标题/表格/公式等场景)](https://github.com/UnicomAI/DocParserServer/tree/main)的私有化部署及接入，文档分段设置支持通用分段和父子分段。在调优能力方面:支持知识图谱、元数据管理及元数据过滤查询，支持分段内容增删改，支持对分段设置关键词标签提升召回效果，支持分段启停操作，支持命中测试等功能。在检索能力方面:支持向量检索、全文检索、混合检索多种检索模式;在问答能力方面:支持自动引用出处，支持图文并茂的生成答案。 |
 | [资源库](https://github.com/UnicomAI/wanwu/blob/main/configs/microservice/bff-service/static/manual/3.%E5%B7%A5%E5%85%B7%E5%B9%BF%E5%9C%BA.md) | 同时支持导入自己的MCP服务或自定义工具或提示词，并在工作流和智能体中使用；支持用户创建MCP Server |
 | [安全护栏](https://github.com/UnicomAI/wanwu/blob/main/configs/microservice/bff-service/static/manual/4.%E5%AE%89%E5%85%A8%E6%8A%A4%E6%A0%8F.md) |        用户可以创建敏感词表，控制模型反馈结果的安全性        |

diff --git a/pkg/model-provider/common.go b/pkg/model-provider/common.go
@@ -27,6 +27,7 @@ const (
 	ProviderQianfan          = "QianFan"
 	ProviderDeepSeek         = "DeepSeek"
 	ProviderJina             = "Jina"
+	ProviderMiniMax          = "MiniMax"
 )
 
 var (

diff --git a/pkg/model-provider/model_config_api.go b/pkg/model-provider/model_config_api.go
@@ -10,6 +10,7 @@ import (
 	mp_huoshan "github.com/UnicomAI/wanwu/pkg/model-provider/mp-huoshan"
 	mp_infini "github.com/UnicomAI/wanwu/pkg/model-provider/mp-infini"
 	mp_jina "github.com/UnicomAI/wanwu/pkg/model-provider/mp-jina"
+	mp_minimax "github.com/UnicomAI/wanwu/pkg/model-provider/mp-minimax"
 	mp_ollama "github.com/UnicomAI/wanwu/pkg/model-provider/mp-ollama"
 	mp_openai_compatible "github.com/UnicomAI/wanwu/pkg/model-provider/mp-openai-compatible"
 	mp_qianfan "github.com/UnicomAI/wanwu/pkg/model-provider/mp-qianfan"
@@ -322,6 +323,17 @@ func ToModelTags(provider, modelType, cfg string) ([]mp_common.Tag, error) {
 		default:
 			return nil, fmt.Errorf("ToModelTags:invalid provider %v model type %v", provider, modelType)
 		}
+	case ProviderMiniMax:
+		switch modelType {
+		case ModelTypeLLM:
+			llm := &mp_minimax.LLM{}
+			if err := json.Unmarshal([]byte(cfg), llm); err != nil {
+				return nil, fmt.Errorf("unmarshal model config err: %v", err)
+			}
+			tags = llm.Tags()
+		default:
+			return nil, fmt.Errorf("ToModelTags:invalid provider %v model type %v", provider, modelType)
+		}
 	default:
 		return nil, fmt.Errorf("ToModelTags:invalid provider: %v", provider)
 	}
@@ -470,6 +482,13 @@ func ToModelConfig(provider, modelType, cfg string) (interface{}, error) {
 		default:
 			return nil, fmt.Errorf("ToModelConfig:invalid provider %v model type %v", provider, modelType)
 		}
+	case ProviderMiniMax:
+		switch modelType {
+		case ModelTypeLLM:
+			ret = &mp_minimax.LLM{}
+		default:
+			return nil, fmt.Errorf("ToModelConfig:invalid provider %v model type %v", provider, modelType)
+		}
 	default:
 		return nil, fmt.Errorf("ToModelConfig:invalid provider: %v", provider)
 	}
@@ -489,6 +508,7 @@ type ProviderModelConfig struct {
 	ProviderInfini           ProviderModelByInfini           `json:"providerInfini"`
 	ProviderQianFan          ProviderModelByQianFan          `json:"providerQianFan"`
 	ProviderDeepSeek         ProviderModelByDeepSeek         `json:"providerDeepSeek"`
+	ProviderMiniMax          ProviderModelByMiniMax          `json:"providerMiniMax"`
 }
 
 type ProviderModelByOpenAICompatible struct {
@@ -537,3 +557,7 @@ type ProviderModelByQianFan struct {
 type ProviderModelByDeepSeek struct {
 	Llm mp_deepseek.LLM `json:"llm"`
 }
+
+type ProviderModelByMiniMax struct {
+	Llm mp_minimax.LLM `json:"llm"`
+}
diff --git a/pkg/model-provider/model_params_api.go b/pkg/model-provider/model_params_api.go
@@ -141,6 +141,17 @@ func ToModelParams(provider, modelType, cfg string) (interface{}, map[string]int
 		default:
 			return nil, nil, fmt.Errorf("invalid model type: %v", modelType)
 		}
+	case ProviderMiniMax:
+		switch modelType {
+		case ModelTypeLLM:
+			llm := &mp_common.LLMParams{}
+			if err = json.Unmarshal([]byte(cfg), llm); err == nil {
+				ret = llm
+				params = llm.GetParams()
+			}
+		default:
+			return nil, nil, fmt.Errorf("invalid model type: %v", modelType)
+		}
 	default:
 		return nil, nil, fmt.Errorf("invalid provider: %v", modelType)
 	}
@@ -159,6 +170,7 @@ type AppModelParams struct {
 	ProviderInfini           AppModelParamsInfini           `json:"providerInfini"`
 	ProviderQianFan          AppModelParamsQianFan          `json:"providerQianFan"`
 	ProviderDeepSeek         AppModelParamsDeepSeek         `json:"providerDeepSeek"`
+	ProviderMiniMax          AppModelParamsMiniMax          `json:"providerMiniMax"` // MiniMax模型配置
 }
 
 type AppModelParamsOpenAICompatible struct {
@@ -192,3 +204,7 @@ type AppModelParamsQianFan struct {
 type AppModelParamsDeepSeek struct {
 	LLM mp_common.LLMParams `json:"llm"` // 大语言模型配置
 }
+
+type AppModelParamsMiniMax struct {
+	LLM mp_common.LLMParams `json:"llm"` // 大语言模型配置
+}
diff --git a/pkg/model-provider/mp-minimax/mp-minimax-llm.go b/pkg/model-provider/mp-minimax/mp-minimax-llm.go
@@ -0,0 +1,64 @@
+package mp_minimax
+
+import (
+	"context"
+	"fmt"
+	"net/url"
+
+	mp_common "github.com/UnicomAI/wanwu/pkg/model-provider/mp-common"
+)
+
+type LLM struct {
+	ApiKey          string `json:"apiKey"`                                              // ApiKey
+	EndpointUrl     string `json:"endpointUrl"`                                         // Inference URL (default: https://api.minimax.io/v1)
+	FunctionCalling string `json:"functionCalling" validate:"oneof=noSupport toolCall"` // Function calling support
+	VisionSupport   string `json:"visionSupport" validate:"oneof=noSupport support"`    // Vision support
+	ThinkingSupport string `json:"thinkingSupport" validate:"oneof=noSupport support"`  // Deep thinking support
+	MaxTokens       *int   `json:"maxTokens"`                                           // Max output tokens
+	ContextSize     *int   `json:"contextSize"`                                         // Context window size
+	MaxImageSize    *int64 `json:"maxImageSize"`                                        // Max image size limit
+}
+
+func (cfg *LLM) Tags() []mp_common.Tag {
+	tags := []mp_common.Tag{
+		{
+			Text: mp_common.TagChat,
+		},
+	}
+	tags = append(tags, mp_common.GetTagsByFunctionCall(cfg.FunctionCalling)...)
+	tags = append(tags, mp_common.GetTagsByVisionSupport(cfg.VisionSupport)...)
+	tags = append(tags, mp_common.GetTagsByContentSize(cfg.ContextSize)...)
+	return tags
+}
+
+func (cfg *LLM) NewReq(req *mp_common.LLMReq) (mp_common.ILLMReq, error) {
+	if req.MaxTokens != nil && cfg.ContextSize != nil && *req.MaxTokens > *cfg.ContextSize {
+		return nil, fmt.Errorf("max_tokens too large (max allowed: %d)", *cfg.ContextSize)
+	}
+	m, err := req.Data()
+	if err != nil {
+		return nil, err
+	}
+	// MiniMax temperature range is [0, 1.0]; clamp if needed
+	if req.Temperature != nil && *req.Temperature > 1.0 {
+		clamped := 1.0
+		m["temperature"] = clamped
+	}
+	if req.Stream != nil && *req.Stream {
+		if req.StreamOptions != nil && req.StreamOptions.IncludeUsage != nil {
+			m["stream_options"] = map[string]bool{"include_usage": *req.StreamOptions.IncludeUsage}
+		} else {
+			m["stream_options"] = map[string]bool{"include_usage": true}
+		}
+	}
+	return mp_common.NewLLMReq(m), nil
+}
+
+func (cfg *LLM) ChatCompletions(ctx context.Context, req mp_common.ILLMReq, headers ...mp_common.Header) (mp_common.ILLMResp, <-chan mp_common.ILLMResp, error) {
+	return mp_common.ChatCompletions(ctx, "minimax", cfg.ApiKey, cfg.chatCompletionsUrl(), req, mp_common.NewLLMResp, headers...)
+}
+
+func (cfg *LLM) chatCompletionsUrl() string {
+	ret, _ := url.JoinPath(cfg.EndpointUrl, "/chat/completions")
+	return ret
+}