feat(model):租户模型定价和显示优化

api/admin/v1/tenant_model.go

@@ -13,21 +13,34 @@ type TenantModelListRes struct {
 }
 
 type TenantModelItem struct {
-	ID                int64    `json:"id"`
-	TenantID          int64    `json:"tenant_id"`
-	ModelID           int64    `json:"model_id"`
-	ModelCode         string   `json:"model_code"`
-	ModelName         string   `json:"model_name"`
-	Category          string   `json:"category"`
-	Enabled           bool     `json:"enabled"`
-	BillingMode       *string  `json:"billing_mode"`
-	PerRequestPrice   *float64 `json:"per_request_price"`
-	DiscountRatio     *float64 `json:"discount_ratio"`
-	MaxConcurrency    *int     `json:"max_concurrency"`
-	ChannelScope      string   `json:"channel_scope"`
-	CustomInputPrice  *float64 `json:"custom_input_price"`
-	CustomOutputPrice *float64 `json:"custom_output_price"`
-	Multiplier        float64  `json:"multiplier"`
+	ID                       int64          `json:"id"`
+	TenantID                 int64          `json:"tenant_id"`
+	ModelID                  int64          `json:"model_id"`
+	ModelCode                string         `json:"model_code"`
+	ModelName                string         `json:"model_name"`
+	Category                 string         `json:"category"`
+	Enabled                  bool           `json:"enabled"`
+	BillingMode              *string        `json:"billing_mode"`
+	PerRequestPrice          *float64       `json:"per_request_price"`
+	DiscountRatio            *float64       `json:"discount_ratio"`
+	MaxConcurrency           *int           `json:"max_concurrency"`
+	ChannelScope             string         `json:"channel_scope"`
+	CustomInputPrice         *float64       `json:"custom_input_price"`
+	CustomOutputPrice        *float64       `json:"custom_output_price"`
+	CustomCacheReadPrice     *float64       `json:"custom_cache_read_price"`
+	CustomCacheCreationPrice *float64       `json:"custom_cache_creation_price"`
+	CustomPricingTiers       []*PricingTier `json:"custom_pricing_tiers"`
+	Multiplier               float64        `json:"multiplier"`
+}
+
+// PricingTier 阶梯定价项（租户自定义阶梯时使用）
+type PricingTier struct {
+	MinTokens          int64   `json:"min_tokens"`
+	MaxTokens          *int64  `json:"max_tokens"`
+	InputPrice         float64 `json:"input_price"`
+	OutputPrice        float64 `json:"output_price"`
+	CacheReadPrice     float64 `json:"cache_read_price"`
+	CacheCreationPrice float64 `json:"cache_creation_price"`
 }
 
 // TenantModelBatchAssignReq 批量分配模型给租户
@@ -38,15 +51,18 @@ type TenantModelBatchAssignReq struct {
 }
 
 type ModelAssignment struct {
-	ModelID           int64    `json:"model_id" v:"required" dc:"模型ID"`
-	Enabled           bool     `json:"enabled" d:"true" dc:"是否启用"`
-	BillingMode       *string  `json:"billing_mode" dc:"覆盖计费模式"`
-	PerRequestPrice   *float64 `json:"per_request_price" dc:"按次单价"`
-	DiscountRatio     *float64 `json:"discount_ratio" dc:"折扣比例(0-1)"`
-	MaxConcurrency    *int     `json:"max_concurrency" dc:"单模型并发上限"`
-	ChannelScope      *string  `json:"channel_scope" dc:"渠道范围JSON数组"`
-	CustomInputPrice  *float64 `json:"custom_input_price" dc:"自定义输入价格"`
-	CustomOutputPrice *float64 `json:"custom_output_price" dc:"自定义输出价格"`
+	ModelID                  int64          `json:"model_id" v:"required" dc:"模型ID"`
+	Enabled                  bool           `json:"enabled" d:"true" dc:"是否启用"`
+	BillingMode              *string        `json:"billing_mode" dc:"覆盖计费模式"`
+	PerRequestPrice          *float64       `json:"per_request_price" dc:"按次单价"`
+	DiscountRatio            *float64       `json:"discount_ratio" dc:"折扣比例(0-1)"`
+	MaxConcurrency           *int           `json:"max_concurrency" dc:"单模型并发上限"`
+	ChannelScope             *string        `json:"channel_scope" dc:"渠道范围JSON数组"`
+	CustomInputPrice         *float64       `json:"custom_input_price" dc:"自定义输入价格"`
+	CustomOutputPrice        *float64       `json:"custom_output_price" dc:"自定义输出价格"`
+	CustomCacheReadPrice     *float64       `json:"custom_cache_read_price" dc:"自定义缓存读取价格"`
+	CustomCacheCreationPrice *float64       `json:"custom_cache_creation_price" dc:"自定义缓存创建价格"`
+	CustomPricingTiers       []*PricingTier `json:"custom_pricing_tiers" dc:"自定义阶梯定价"`
 }
 
 type TenantModelBatchAssignRes struct {
@@ -55,17 +71,20 @@ type TenantModelBatchAssignRes struct {
 
 // TenantModelUpdateReq 更新租户模型分配配置
 type TenantModelUpdateReq struct {
-	g.Meta            `path:"/tenants/{tenant_id}/models/{model_id}" method:"put" mime:"json" tags:"管理后台-租户模型" summary:"更新租户模型配置"`
-	TenantID          int64     `json:"tenant_id" in:"path" v:"required" dc:"租户ID"`
-	ModelID           int64     `json:"model_id" in:"path" v:"required" dc:"模型ID"`
-	Enabled           *bool     `json:"enabled" dc:"是否启用"`
-	BillingMode       *string   `json:"billing_mode" v:"in:token,per_request" dc:"计费模式"`
-	PerRequestPrice   **float64 `json:"per_request_price" dc:"按次单价"`
-	DiscountRatio     **float64 `json:"discount_ratio" dc:"折扣比例"`
-	MaxConcurrency    **int     `json:"max_concurrency" dc:"单模型并发上限"`
-	ChannelScope      **string  `json:"channel_scope" dc:"渠道范围JSON数组"`
-	CustomInputPrice  **float64 `json:"custom_input_price" dc:"自定义输入价格"`
-	CustomOutputPrice **float64 `json:"custom_output_price" dc:"自定义输出价格"`
+	g.Meta                   `path:"/tenants/{tenant_id}/models/{model_id}" method:"put" mime:"json" tags:"管理后台-租户模型" summary:"更新租户模型配置"`
+	TenantID                 int64           `json:"tenant_id" in:"path" v:"required" dc:"租户ID"`
+	ModelID                  int64           `json:"model_id" in:"path" v:"required" dc:"模型ID"`
+	Enabled                  *bool           `json:"enabled" dc:"是否启用"`
+	BillingMode              *string         `json:"billing_mode" v:"in:token,per_request,tiered" dc:"计费模式"`
+	PerRequestPrice          **float64       `json:"per_request_price" dc:"按次单价"`
+	DiscountRatio            **float64       `json:"discount_ratio" dc:"折扣比例"`
+	MaxConcurrency           **int           `json:"max_concurrency" dc:"单模型并发上限"`
+	ChannelScope             **string        `json:"channel_scope" dc:"渠道范围JSON数组"`
+	CustomInputPrice         **float64       `json:"custom_input_price" dc:"自定义输入价格"`
+	CustomOutputPrice        **float64       `json:"custom_output_price" dc:"自定义输出价格"`
+	CustomCacheReadPrice     **float64       `json:"custom_cache_read_price" dc:"自定义缓存读取价格"`
+	CustomCacheCreationPrice **float64       `json:"custom_cache_creation_price" dc:"自定义缓存创建价格"`
+	CustomPricingTiers       *[]*PricingTier `json:"custom_pricing_tiers" dc:"自定义阶梯定价"`
 }
 
 type TenantModelUpdateRes struct{}

api/tenant/v1/models.go

@@ -13,19 +13,32 @@ type TenantAvailableModelsRes struct {
 	List []TenantAvailableModelItem `json:"list"`
 }
 
+// PricingTierItem 阶梯定价项
+type PricingTierItem struct {
+	MinTokens   int64   `json:"min_tokens"`
+	MaxTokens   *int64  `json:"max_tokens"`
+	InputPrice  float64 `json:"input_price"`
+	OutputPrice float64 `json:"output_price"`
+}
+
 // TenantAvailableModelItem 租户可用模型信息
 type TenantAvailableModelItem struct {
-	ID              int64    `json:"id"`
-	ModelId         string   `json:"model_id"`
-	ModelName       string   `json:"model_name"`
-	Category        string   `json:"category"`
-	MaxContext      int      `json:"max_context_tokens"`
-	MaxOutput       int      `json:"max_output_tokens"`
-	Description     string   `json:"description"`
-	Tags            string   `json:"tags"`
-	Capabilities    string   `json:"capabilities"`
-	BillingMode     *string  `json:"billing_mode"`
-	PerRequestPrice *float64 `json:"per_request_price"`
-	DiscountRatio   *float64 `json:"discount_ratio"`
-	MaxConcurrency  *int     `json:"max_concurrency"`
+	ID                 int64             `json:"id"`
+	ModelId            string            `json:"model_id"`
+	ModelName          string            `json:"model_name"`
+	Category           string            `json:"category"`
+	MaxContext         int               `json:"max_context_tokens"`
+	MaxOutput          int               `json:"max_output_tokens"`
+	Description        string            `json:"description"`
+	Tags               string            `json:"tags"`
+	Capabilities       string            `json:"capabilities"`
+	BillingMode        *string           `json:"billing_mode"`
+	PerRequestPrice    *float64          `json:"per_request_price"`
+	DiscountRatio      *float64          `json:"discount_ratio"`
+	MaxConcurrency     *int              `json:"max_concurrency"`
+	InputPrice         *float64          `json:"input_price"`
+	OutputPrice        *float64          `json:"output_price"`
+	CacheReadPrice     *float64          `json:"cache_read_price"`
+	CacheCreationPrice *float64          `json:"cache_creation_price"`
+	PricingTiers       []PricingTierItem `json:"pricing_tiers"`
 }

internal/dao/internal/chn_channel_keys.go

@@ -29,9 +29,9 @@ type ChnChannelKeysColumns struct {
 	LastUsedAt     string // 最后使用时间
 	LastError      string // 最后一次错误信息
 	CreatedAt      string // 创建时间
-	UpdatedAt      string // 更新时间
 	KeyType        string // Key 类型：apikey（传统静态密钥）/ oauth（OAuth 令牌）
 	TokenExpiresAt string // OAuth access_token 过期时间（仅 key_type=oauth 时有值）
+	UpdatedAt      string // 更新时间
 }
 
 // chnChannelKeysColumns holds the columns for the table chn_channel_keys.
@@ -44,9 +44,9 @@ var chnChannelKeysColumns = ChnChannelKeysColumns{
 	LastUsedAt:     "last_used_at",
 	LastError:      "last_error",
 	CreatedAt:      "created_at",
-	UpdatedAt:      "updated_at",
 	KeyType:        "key_type",
 	TokenExpiresAt: "token_expires_at",
+	UpdatedAt:      "updated_at",
 }
 
 // NewChnChannelKeysDao creates and returns a new DAO object for table data access.

internal/dao/internal/mdl_models.go

@@ -30,12 +30,12 @@ type MdlModelsColumns struct {
 	MaxOutputTokens  string // 最大输出 token 数
 	Description      string // 模型描述
 	Tags             string // 标签（如 reasoning、vision、function_calling）
+	Capabilities     string // 模型能力特性（如 vision、function_calling、reasoning 等）
 	CreatedAt        string // 创建时间
 	UpdatedAt        string // 更新时间
 	DeprecatedAt     string // 标记弃用的时间（NULL表示未弃用）
 	SunsetDate       string // 计划下线日期（到达后返回410 Gone，NULL表示未设置）
 	ReplacementModel string // 推荐替代模型名（NULL表示无替代）
-	Capabilities     string // 模型能力特性（如 vision、function_calling、reasoning 等）
 }
 
 // mdlModelsColumns holds the columns for the table mdl_models.
@@ -49,12 +49,12 @@ var mdlModelsColumns = MdlModelsColumns{
 	MaxOutputTokens:  "max_output_tokens",
 	Description:      "description",
 	Tags:             "tags",
+	Capabilities:     "capabilities",
 	CreatedAt:        "created_at",
 	UpdatedAt:        "updated_at",
 	DeprecatedAt:     "deprecated_at",
 	SunsetDate:       "sunset_date",
 	ReplacementModel: "replacement_model",
-	Capabilities:     "capabilities",
 }
 
 // NewMdlModelsDao creates and returns a new DAO object for table data access.

internal/dao/internal/mdl_tenant_models.go

@@ -21,38 +21,44 @@ type MdlTenantModelsDao struct {
 
 // MdlTenantModelsColumns defines and stores column names for the table mdl_tenant_models.
 type MdlTenantModelsColumns struct {
-	Id                string // 主键ID
-	TenantId          string // 租户ID
-	ModelId           string // 模型ID
-	Enabled           string // 是否启用（禁用后该租户无法调用此模型）
-	CustomInputPrice  string // 租户自定义输入价格（NULL 表示使用默认定价）
-	CustomOutputPrice string // 租户自定义输出价格（NULL 表示使用默认定价）
-	Multiplier        string // 租户价格倍率（VIP 折扣等，最终价格 = 基础价格 × 模型倍率 × 租户倍率）
-	CreatedAt         string // 创建时间
-	UpdatedAt         string // 更新时间
-	BillingMode       string // 覆盖模型计费方式（NULL表示跟随模型默认）
-	PerRequestPrice   string // 按次计费单价（覆盖模型默认，仅 billing_mode = per_request 时有效）
-	DiscountRatio     string // 折扣比例（如0.8表示八折，NULL表示不打折，优先于 multiplier 使用）
-	MaxConcurrency    string // 单模型并发上限（NULL表示不限制）
-	ChannelScope      string // 渠道范围覆盖（NULL跟随租户默认，[]表示全部，数组表示指定渠道ID）
+	Id                       string // 主键ID
+	TenantId                 string // 租户ID
+	ModelId                  string // 模型ID
+	Enabled                  string // 是否启用（禁用后该租户无法调用此模型）
+	CustomInputPrice         string // 租户自定义输入价格（NULL 表示使用默认定价）
+	CustomOutputPrice        string // 租户自定义输出价格（NULL 表示使用默认定价）
+	Multiplier               string // 租户价格倍率（VIP 折扣等，最终价格 = 基础价格 × 模型倍率 × 租户倍率）
+	CreatedAt                string // 创建时间
+	UpdatedAt                string // 更新时间
+	BillingMode              string // 覆盖模型计费方式（NULL表示跟随模型默认）
+	PerRequestPrice          string // 按次计费单价（覆盖模型默认，仅 billing_mode = per_request 时有效）
+	DiscountRatio            string // 折扣比例（如0.8表示八折，NULL表示不打折，优先于 multiplier 使用）
+	MaxConcurrency           string // 单模型并发上限（NULL表示不限制）
+	ChannelScope             string // 渠道范围覆盖（NULL跟随租户默认，[]表示全部，数组表示指定渠道ID）
+	CustomCacheReadPrice     string // 自定义缓存读取价格（$/1M token），NULL 表示使用基础定价
+	CustomCacheCreationPrice string // 自定义缓存创建价格（$/1M token），NULL 表示使用基础定价
+	CustomPricingTiers       string // 自定义阶梯定价（JSONB 数组），格式: [{"min_tokens":0,"max_tokens":100000,"input_price":0.5,"output_price":1.5,"cache_read_price":0.1,"cache_creation_price":0.2}]
 }
 
 // mdlTenantModelsColumns holds the columns for the table mdl_tenant_models.
 var mdlTenantModelsColumns = MdlTenantModelsColumns{
-	Id:                "id",
-	TenantId:          "tenant_id",
-	ModelId:           "model_id",
-	Enabled:           "enabled",
-	CustomInputPrice:  "custom_input_price",
-	CustomOutputPrice: "custom_output_price",
-	Multiplier:        "multiplier",
-	CreatedAt:         "created_at",
-	UpdatedAt:         "updated_at",
-	BillingMode:       "billing_mode",
-	PerRequestPrice:   "per_request_price",
-	DiscountRatio:     "discount_ratio",
-	MaxConcurrency:    "max_concurrency",
-	ChannelScope:      "channel_scope",
+	Id:                       "id",
+	TenantId:                 "tenant_id",
+	ModelId:                  "model_id",
+	Enabled:                  "enabled",
+	CustomInputPrice:         "custom_input_price",
+	CustomOutputPrice:        "custom_output_price",
+	Multiplier:               "multiplier",
+	CreatedAt:                "created_at",
+	UpdatedAt:                "updated_at",
+	BillingMode:              "billing_mode",
+	PerRequestPrice:          "per_request_price",
+	DiscountRatio:            "discount_ratio",
+	MaxConcurrency:           "max_concurrency",
+	ChannelScope:             "channel_scope",
+	CustomCacheReadPrice:     "custom_cache_read_price",
+	CustomCacheCreationPrice: "custom_cache_creation_price",
+	CustomPricingTiers:       "custom_pricing_tiers",
 }
 
 // NewMdlTenantModelsDao creates and returns a new DAO object for table data access.

internal/dao/internal/ntf_messages.go

@@ -31,8 +31,8 @@ type NtfMessagesColumns struct {
 	IsRead      string // 是否已读：0=未读, 1=已读
 	IsBroadcast string // 是否广播消息：0=个人消息, 1=广播消息
 	Metadata    string // 附加元数据（JSONB，如关联资源ID、跳转链接等）
-	CreatedAt   string // 创建时间
 	TargetRoles string // 目标角色（NULL=全部角色，逗号分隔如 owner,admin 表示仅限这些角色）
+	CreatedAt   string // 创建时间
 }
 
 // ntfMessagesColumns holds the columns for the table ntf_messages.
@@ -47,8 +47,8 @@ var ntfMessagesColumns = NtfMessagesColumns{
 	IsRead:      "is_read",
 	IsBroadcast: "is_broadcast",
 	Metadata:    "metadata",
-	CreatedAt:   "created_at",
 	TargetRoles: "target_roles",
+	CreatedAt:   "created_at",
 }
 
 // NewNtfMessagesDao creates and returns a new DAO object for table data access.

internal/dao/internal/sys_plugins.go

@@ -0,0 +1,97 @@
+// ==========================================================================
+// Code generated and maintained by GoFrame CLI tool. DO NOT EDIT.
+// ==========================================================================
+
+package internal
+
+import (
+	"context"
+
+	"github.com/gogf/gf/v2/database/gdb"
+	"github.com/gogf/gf/v2/frame/g"
+)
+
+// SysPluginsDao is the data access object for the table sys_plugins.
+type SysPluginsDao struct {
+	table    string             // table is the underlying table name of the DAO.
+	group    string             // group is the database configuration group name of the current DAO.
+	columns  SysPluginsColumns  // columns contains all the column names of Table for convenient usage.
+	handlers []gdb.ModelHandler // handlers for customized model modification.
+}
+
+// SysPluginsColumns defines and stores column names for the table sys_plugins.
+type SysPluginsColumns struct {
+	Id        string //
+	Name      string // 插件唯一标识，如 email-report
+	Label     string // 显示名称
+	Version   string // 当前安装版本
+	Status    string // 状态：registered=已注册, installed=已安装, enabled=已启用, disabled=已禁用, error=异常
+	Category  string // 分类：relay=代理扩展, middleware=中间件, billing=计费, notification=通知, extension=通用扩展
+	Config    string // 插件全局配置（JSON）
+	ErrorMsg  string // 异常信息
+	CreatedAt string //
+	UpdatedAt string //
+}
+
+// sysPluginsColumns holds the columns for the table sys_plugins.
+var sysPluginsColumns = SysPluginsColumns{
+	Id:        "id",
+	Name:      "name",
+	Label:     "label",
+	Version:   "version",
+	Status:    "status",
+	Category:  "category",
+	Config:    "config",
+	ErrorMsg:  "error_msg",
+	CreatedAt: "created_at",
+	UpdatedAt: "updated_at",
+}
+
+// NewSysPluginsDao creates and returns a new DAO object for table data access.
+func NewSysPluginsDao(handlers ...gdb.ModelHandler) *SysPluginsDao {
+	return &SysPluginsDao{
+		group:    "default",
+		table:    "sys_plugins",
+		columns:  sysPluginsColumns,
+		handlers: handlers,
+	}
+}
+
+// DB retrieves and returns the underlying raw database management object of the current DAO.
+func (dao *SysPluginsDao) DB() gdb.DB {
+	return g.DB(dao.group)
+}
+
+// Table returns the table name of the current DAO.
+func (dao *SysPluginsDao) Table() string {
+	return dao.table
+}
+
+// Columns returns all column names of the current DAO.
+func (dao *SysPluginsDao) Columns() SysPluginsColumns {
+	return dao.columns
+}
+
+// Group returns the database configuration group name of the current DAO.
+func (dao *SysPluginsDao) Group() string {
+	return dao.group
+}
+
+// Ctx creates and returns a Model for the current DAO. It automatically sets the context for the current operation.
+func (dao *SysPluginsDao) Ctx(ctx context.Context) *gdb.Model {
+	model := dao.DB().Model(dao.table)
+	for _, handler := range dao.handlers {
+		model = handler(model)
+	}
+	return model.Safe().Ctx(ctx)
+}
+
+// Transaction wraps the transaction logic using function f.
+// It rolls back the transaction and returns the error if function f returns a non-nil error.
+// It commits the transaction and returns nil if function f returns nil.
+//
+// Note: Do not commit or roll back the transaction in function f,
+// as it is automatically handled by this function.
+func (dao *SysPluginsDao) Transaction(ctx context.Context, f func(ctx context.Context, tx gdb.TX) error) (err error) {
+	return dao.Ctx(ctx).Transaction(ctx, f)
+}

internal/dao/internal/tnt_tenant_plugins.go

@@ -0,0 +1,91 @@
+// ==========================================================================
+// Code generated and maintained by GoFrame CLI tool. DO NOT EDIT.
+// ==========================================================================
+
+package internal
+
+import (
+	"context"
+
+	"github.com/gogf/gf/v2/database/gdb"
+	"github.com/gogf/gf/v2/frame/g"
+)
+
+// TntTenantPluginsDao is the data access object for the table tnt_tenant_plugins.
+type TntTenantPluginsDao struct {
+	table    string                  // table is the underlying table name of the DAO.
+	group    string                  // group is the database configuration group name of the current DAO.
+	columns  TntTenantPluginsColumns // columns contains all the column names of Table for convenient usage.
+	handlers []gdb.ModelHandler      // handlers for customized model modification.
+}
+
+// TntTenantPluginsColumns defines and stores column names for the table tnt_tenant_plugins.
+type TntTenantPluginsColumns struct {
+	Id         string //
+	TenantId   string //
+	PluginName string // 插件标识
+	Enabled    string // 是否启用
+	Config     string // 租户级配置覆盖（JSON），优先级高于全局配置
+	CreatedAt  string //
+	UpdatedAt  string //
+}
+
+// tntTenantPluginsColumns holds the columns for the table tnt_tenant_plugins.
+var tntTenantPluginsColumns = TntTenantPluginsColumns{
+	Id:         "id",
+	TenantId:   "tenant_id",
+	PluginName: "plugin_name",
+	Enabled:    "enabled",
+	Config:     "config",
+	CreatedAt:  "created_at",
+	UpdatedAt:  "updated_at",
+}
+
+// NewTntTenantPluginsDao creates and returns a new DAO object for table data access.
+func NewTntTenantPluginsDao(handlers ...gdb.ModelHandler) *TntTenantPluginsDao {
+	return &TntTenantPluginsDao{
+		group:    "default",
+		table:    "tnt_tenant_plugins",
+		columns:  tntTenantPluginsColumns,
+		handlers: handlers,
+	}
+}
+
+// DB retrieves and returns the underlying raw database management object of the current DAO.
+func (dao *TntTenantPluginsDao) DB() gdb.DB {
+	return g.DB(dao.group)
+}
+
+// Table returns the table name of the current DAO.
+func (dao *TntTenantPluginsDao) Table() string {
+	return dao.table
+}
+
+// Columns returns all column names of the current DAO.
+func (dao *TntTenantPluginsDao) Columns() TntTenantPluginsColumns {
+	return dao.columns
+}
+
+// Group returns the database configuration group name of the current DAO.
+func (dao *TntTenantPluginsDao) Group() string {
+	return dao.group
+}
+
+// Ctx creates and returns a Model for the current DAO. It automatically sets the context for the current operation.
+func (dao *TntTenantPluginsDao) Ctx(ctx context.Context) *gdb.Model {
+	model := dao.DB().Model(dao.table)
+	for _, handler := range dao.handlers {
+		model = handler(model)
+	}
+	return model.Safe().Ctx(ctx)
+}
+
+// Transaction wraps the transaction logic using function f.
+// It rolls back the transaction and returns the error if function f returns a non-nil error.
+// It commits the transaction and returns nil if function f returns nil.
+//
+// Note: Do not commit or roll back the transaction in function f,
+// as it is automatically handled by this function.
+func (dao *TntTenantPluginsDao) Transaction(ctx context.Context, f func(ctx context.Context, tx gdb.TX) error) (err error) {
+	return dao.Ctx(ctx).Transaction(ctx, f)
+}

internal/dao/sys_plugins.go

@@ -0,0 +1,22 @@
+// =================================================================================
+// This file is auto-generated by the GoFrame CLI tool. You may modify it as needed.
+// =================================================================================
+
+package dao
+
+import (
+	"github.com/qianfree/team-api/internal/dao/internal"
+)
+
+// sysPluginsDao is the data access object for the table sys_plugins.
+// You can define custom methods on it to extend its functionality as needed.
+type sysPluginsDao struct {
+	*internal.SysPluginsDao
+}
+
+var (
+	// SysPlugins is a globally accessible object for table sys_plugins operations.
+	SysPlugins = sysPluginsDao{internal.NewSysPluginsDao()}
+)
+
+// Add your custom methods and functionality below.

internal/dao/tnt_tenant_plugins.go

@@ -0,0 +1,22 @@
+// =================================================================================
+// This file is auto-generated by the GoFrame CLI tool. You may modify it as needed.
+// =================================================================================
+
+package dao
+
+import (
+	"github.com/qianfree/team-api/internal/dao/internal"
+)
+
+// tntTenantPluginsDao is the data access object for the table tnt_tenant_plugins.
+// You can define custom methods on it to extend its functionality as needed.
+type tntTenantPluginsDao struct {
+	*internal.TntTenantPluginsDao
+}
+
+var (
+	// TntTenantPlugins is a globally accessible object for table tnt_tenant_plugins operations.
+	TntTenantPlugins = tntTenantPluginsDao{internal.NewTntTenantPluginsDao()}
+)
+
+// Add your custom methods and functionality below.

internal/logic/admin/tenant_model.go

@@ -2,38 +2,42 @@ package admin
 
 import (
 	"context"
-	do "github.com/qianfree/team-api/internal/model/do"
+	"encoding/json"
 
 	v1 "github.com/qianfree/team-api/api/admin/v1"
 	"github.com/qianfree/team-api/internal/dao"
 	"github.com/qianfree/team-api/internal/logic/billing"
 	"github.com/qianfree/team-api/internal/logic/common"
+	do "github.com/qianfree/team-api/internal/model/do"
 )
 
 // ListTenantModels 列出租户已分配的模型
 func (s *sAdmin) ListTenantModels(ctx context.Context, req *v1.TenantModelListReq) (*v1.TenantModelListRes, error) {
 	var results []struct {
-		ID                int64    `json:"id"`
-		TenantID          int64    `json:"tenant_id"`
-		ModelID           int64    `json:"model_id"`
-		Enabled           bool     `json:"enabled"`
-		BillingMode       *string  `json:"billing_mode"`
-		PerRequestPrice   *float64 `json:"per_request_price"`
-		DiscountRatio     *float64 `json:"discount_ratio"`
-		MaxConcurrency    *int     `json:"max_concurrency"`
-		ChannelScope      string   `json:"channel_scope"`
-		CustomInputPrice  *float64 `json:"custom_input_price"`
-		CustomOutputPrice *float64 `json:"custom_output_price"`
-		Multiplier        float64  `json:"multiplier"`
-		ModelCode         string   `json:"model_code"`
-		ModelName         string   `json:"model_name"`
-		Category          string   `json:"category"`
+		ID                       int64    `json:"id"`
+		TenantID                 int64    `json:"tenant_id"`
+		ModelID                  int64    `json:"model_id"`
+		Enabled                  bool     `json:"enabled"`
+		BillingMode              *string  `json:"billing_mode"`
+		PerRequestPrice          *float64 `json:"per_request_price"`
+		DiscountRatio            *float64 `json:"discount_ratio"`
+		MaxConcurrency           *int     `json:"max_concurrency"`
+		ChannelScope             string   `json:"channel_scope"`
+		CustomInputPrice         *float64 `json:"custom_input_price"`
+		CustomOutputPrice        *float64 `json:"custom_output_price"`
+		CustomCacheReadPrice     *float64 `json:"custom_cache_read_price"`
+		CustomCacheCreationPrice *float64 `json:"custom_cache_creation_price"`
+		CustomPricingTiers       string   `json:"custom_pricing_tiers"`
+		Multiplier               float64  `json:"multiplier"`
+		ModelCode                string   `json:"model_code"`
+		ModelName                string   `json:"model_name"`
+		Category                 string   `json:"category"`
 	}
 
 	err := dao.MdlTenantModels.Ctx(ctx).
 		LeftJoin("mdl_models ON mdl_tenant_models.model_id = mdl_models.id").
 		Where("mdl_tenant_models.tenant_id", req.TenantID).
-		Fields("mdl_tenant_models.id, mdl_tenant_models.tenant_id, mdl_tenant_models.model_id, mdl_tenant_models.enabled, mdl_tenant_models.billing_mode, mdl_tenant_models.per_request_price, mdl_tenant_models.discount_ratio, mdl_tenant_models.max_concurrency, mdl_tenant_models.channel_scope, mdl_tenant_models.custom_input_price, mdl_tenant_models.custom_output_price, mdl_tenant_models.multiplier, mdl_models.model_id as model_code, mdl_models.model_name, mdl_models.category").
+		Fields("mdl_tenant_models.id, mdl_tenant_models.tenant_id, mdl_tenant_models.model_id, mdl_tenant_models.enabled, mdl_tenant_models.billing_mode, mdl_tenant_models.per_request_price, mdl_tenant_models.discount_ratio, mdl_tenant_models.max_concurrency, mdl_tenant_models.channel_scope, mdl_tenant_models.custom_input_price, mdl_tenant_models.custom_output_price, mdl_tenant_models.custom_cache_read_price, mdl_tenant_models.custom_cache_creation_price, mdl_tenant_models.custom_pricing_tiers, mdl_tenant_models.multiplier, mdl_models.model_id as model_code, mdl_models.model_name, mdl_models.category").
 		OrderAsc("mdl_models.category").
 		OrderAsc("mdl_models.model_id").
 		Scan(&results)
@@ -43,23 +47,32 @@ func (s *sAdmin) ListTenantModels(ctx context.Context, req *v1.TenantModelListRe
 
 	list := make([]v1.TenantModelItem, 0, len(results))
 	for _, r := range results {
-		list = append(list, v1.TenantModelItem{
-			ID:                r.ID,
-			TenantID:          r.TenantID,
-			ModelID:           r.ModelID,
-			ModelCode:         r.ModelCode,
-			ModelName:         r.ModelName,
-			Category:          r.Category,
-			Enabled:           r.Enabled,
-			BillingMode:       r.BillingMode,
-			PerRequestPrice:   r.PerRequestPrice,
-			DiscountRatio:     r.DiscountRatio,
-			MaxConcurrency:    r.MaxConcurrency,
-			ChannelScope:      r.ChannelScope,
-			CustomInputPrice:  r.CustomInputPrice,
-			CustomOutputPrice: r.CustomOutputPrice,
-			Multiplier:        r.Multiplier,
-		})
+		item := v1.TenantModelItem{
+			ID:                       r.ID,
+			TenantID:                 r.TenantID,
+			ModelID:                  r.ModelID,
+			ModelCode:                r.ModelCode,
+			ModelName:                r.ModelName,
+			Category:                 r.Category,
+			Enabled:                  r.Enabled,
+			BillingMode:              r.BillingMode,
+			PerRequestPrice:          r.PerRequestPrice,
+			DiscountRatio:            r.DiscountRatio,
+			MaxConcurrency:           r.MaxConcurrency,
+			ChannelScope:             r.ChannelScope,
+			CustomInputPrice:         r.CustomInputPrice,
+			CustomOutputPrice:        r.CustomOutputPrice,
+			CustomCacheReadPrice:     r.CustomCacheReadPrice,
+			CustomCacheCreationPrice: r.CustomCacheCreationPrice,
+			Multiplier:               r.Multiplier,
+		}
+		if r.CustomPricingTiers != "" && r.CustomPricingTiers != "null" && r.CustomPricingTiers != "[]" {
+			var tiers []*v1.PricingTier
+			if err := json.Unmarshal([]byte(r.CustomPricingTiers), &tiers); err == nil {
+				item.CustomPricingTiers = tiers
+			}
+		}
+		list = append(list, item)
 	}
 
 	return &v1.TenantModelListRes{List: list}, nil
@@ -70,7 +83,6 @@ func (s *sAdmin) BatchAssignModels(ctx context.Context, req *v1.TenantModelBatch
 	assigned := 0
 
 	for _, a := range req.Assignments {
-		// Check model exists and is active
 		var model *struct {
 			ID     int64  `json:"id"`
 			Status string `json:"status"`
@@ -82,7 +94,6 @@ func (s *sAdmin) BatchAssignModels(ctx context.Context, req *v1.TenantModelBatch
 			continue
 		}
 
-		// Check if already assigned
 		count, _ := dao.MdlTenantModels.Ctx(ctx).
 			Where("tenant_id", req.TenantID).
 			Where("model_id", a.ModelID).
@@ -97,25 +108,33 @@ func (s *sAdmin) BatchAssignModels(ctx context.Context, req *v1.TenantModelBatch
 			maxConc = &concurrency
 		}
 
+		var tiersJSON string
+		if len(a.CustomPricingTiers) > 0 {
+			b, _ := json.Marshal(a.CustomPricingTiers)
+			tiersJSON = string(b)
+		}
+
 		_, err = dao.MdlTenantModels.Ctx(ctx).Insert(do.MdlTenantModels{
-			TenantId:          req.TenantID,
-			ModelId:           a.ModelID,
-			Enabled:           a.Enabled,
-			BillingMode:       a.BillingMode,
-			PerRequestPrice:   a.PerRequestPrice,
-			DiscountRatio:     a.DiscountRatio,
-			MaxConcurrency:    maxConc,
-			ChannelScope:      a.ChannelScope,
-			CustomInputPrice:  a.CustomInputPrice,
-			CustomOutputPrice: a.CustomOutputPrice,
-			Multiplier:        1.0,
+			TenantId:                 req.TenantID,
+			ModelId:                  a.ModelID,
+			Enabled:                  a.Enabled,
+			BillingMode:              a.BillingMode,
+			PerRequestPrice:          a.PerRequestPrice,
+			DiscountRatio:            a.DiscountRatio,
+			MaxConcurrency:           maxConc,
+			ChannelScope:             a.ChannelScope,
+			CustomInputPrice:         a.CustomInputPrice,
+			CustomOutputPrice:        a.CustomOutputPrice,
+			CustomCacheReadPrice:     a.CustomCacheReadPrice,
+			CustomCacheCreationPrice: a.CustomCacheCreationPrice,
+			CustomPricingTiers:       tiersJSON,
+			Multiplier:               1.0,
 		})
 		if err == nil {
 			assigned++
 		}
 	}
 
-	// Clear billing cache for this tenant
 	billing.ClearTenantPriceCache(ctx, req.TenantID)
 
 	return &v1.TenantModelBatchAssignRes{Assigned: assigned}, nil
@@ -146,9 +165,23 @@ func (s *sAdmin) UpdateTenantModel(ctx context.Context, req *v1.TenantModelUpdat
 	if req.CustomInputPrice != nil {
 		data.CustomInputPrice = *req.CustomInputPrice
 	}
+	if req.CustomCacheReadPrice != nil {
+		data.CustomCacheReadPrice = *req.CustomCacheReadPrice
+	}
+	if req.CustomCacheCreationPrice != nil {
+		data.CustomCacheCreationPrice = *req.CustomCacheCreationPrice
+	}
 	if req.CustomOutputPrice != nil {
 		data.CustomOutputPrice = *req.CustomOutputPrice
 	}
+	if req.CustomPricingTiers != nil {
+		if len(*req.CustomPricingTiers) > 0 {
+			b, _ := json.Marshal(*req.CustomPricingTiers)
+			data.CustomPricingTiers = string(b)
+		} else {
+			data.CustomPricingTiers = "[]"
+		}
+	}
 
 	_, err := dao.MdlTenantModels.Ctx(ctx).
 		Where("tenant_id", req.TenantID).
@@ -159,7 +192,6 @@ func (s *sAdmin) UpdateTenantModel(ctx context.Context, req *v1.TenantModelUpdat
 		return nil, err
 	}
 
-	// Clear billing cache
 	billing.ClearTenantPriceCache(ctx, req.TenantID)
 
 	return nil, nil
@@ -180,7 +212,6 @@ func (s *sAdmin) DeleteTenantModel(ctx context.Context, req *v1.TenantModelDelet
 		return nil, common.NewNotFoundError("模型分配")
 	}
 
-	// Clear billing cache
 	billing.ClearTenantPriceCache(ctx, req.TenantID)
 
 	return nil, nil

internal/logic/billing/pricing.go

@@ -2,6 +2,7 @@ package billing
 
 import (
 	"context"
+	"encoding/json"
 	"fmt"
 	"math"
 	"time"
@@ -45,6 +46,9 @@ type PricingResult struct {
 	// Cache 直接定价
 	CacheReadPrice     float64 // 缓存读取每 1M token 价格
 	CacheCreationPrice float64 // 缓存创建每 1M token 价格
+
+	// 租户自定义阶梯定价（JSONB 解析后的原始数据，供 CalculateCost 使用）
+	CustomTiers []pricingTierRow
 }
 
 // ClearTenantPriceCache 清除租户的所有模型价格缓存
@@ -139,20 +143,23 @@ func GetModelPrice(ctx context.Context, tenantID int64, modelName string) (*Pric
 
 	// 3. 查租户独立价格（mdl_tenant_models）
 	type tenantModelRow struct {
-		CustomInputPrice  *float64 `json:"custom_input_price"`
-		CustomOutputPrice *float64 `json:"custom_output_price"`
-		Multiplier        *float64 `json:"multiplier"`
-		DiscountRatio     *float64 `json:"discount_ratio"`
-		BillingMode       *string  `json:"billing_mode"`
-		PerRequestPrice   *float64 `json:"per_request_price"`
-		Enabled           bool     `json:"enabled"`
+		CustomInputPrice         *float64 `json:"custom_input_price"`
+		CustomOutputPrice        *float64 `json:"custom_output_price"`
+		CustomCacheReadPrice     *float64 `json:"custom_cache_read_price"`
+		CustomCacheCreationPrice *float64 `json:"custom_cache_creation_price"`
+		CustomPricingTiers       string   `json:"custom_pricing_tiers"`
+		Multiplier               *float64 `json:"multiplier"`
+		DiscountRatio            *float64 `json:"discount_ratio"`
+		BillingMode              *string  `json:"billing_mode"`
+		PerRequestPrice          *float64 `json:"per_request_price"`
+		Enabled                  bool     `json:"enabled"`
 	}
 
 	var tm *tenantModelRow
 	err = dao.MdlTenantModels.Ctx(ctx).
 		Where("tenant_id", tenantID).
 		Where("model_id", model.ID).
-		Fields("custom_input_price, custom_output_price, multiplier, discount_ratio, billing_mode, per_request_price, enabled").
+		Fields("custom_input_price, custom_output_price, custom_cache_read_price, custom_cache_creation_price, custom_pricing_tiers, multiplier, discount_ratio, billing_mode, per_request_price, enabled").
 		Scan(&tm)
 	if err != nil {
 		return nil, gerror.Wrapf(err, "query tenant model price")
@@ -161,6 +168,7 @@ func GetModelPrice(ctx context.Context, tenantID int64, modelName string) (*Pric
 	tenantMultiplier := 1.0
 	discountRatio := 1.0
 	billingSource := "base"
+	var customTiers []pricingTierRow
 
 	if tm != nil && tm.Enabled {
 		billingSource = "tenant_custom"
@@ -176,6 +184,19 @@ func GetModelPrice(ctx context.Context, tenantID int64, modelName string) (*Pric
 			outputPrice = *tm.CustomOutputPrice
 		}
 
+		// 租户覆盖缓存定价
+		if tm.CustomCacheReadPrice != nil && *tm.CustomCacheReadPrice > 0 {
+			cacheReadPrice = *tm.CustomCacheReadPrice
+		}
+		if tm.CustomCacheCreationPrice != nil && *tm.CustomCacheCreationPrice > 0 {
+			cacheCreationPrice = *tm.CustomCacheCreationPrice
+		}
+
+		// 租户自定义阶梯定价
+		if tm.CustomPricingTiers != "" && tm.CustomPricingTiers != "null" && tm.CustomPricingTiers != "[]" {
+			_ = json.Unmarshal([]byte(tm.CustomPricingTiers), &customTiers)
+		}
+
 		// discount_ratio 优先于 multiplier
 		if tm.DiscountRatio != nil && *tm.DiscountRatio > 0 {
 			discountRatio = *tm.DiscountRatio
@@ -211,6 +232,7 @@ func GetModelPrice(ctx context.Context, tenantID int64, modelName string) (*Pric
 		Currency:           "USD",
 		CacheReadPrice:     cacheReadPrice,
 		CacheCreationPrice: cacheCreationPrice,
+		CustomTiers:        customTiers,
 	}
 
 	modelPriceCache.Set(ctx, cacheKey, result)
@@ -245,8 +267,13 @@ func CalculateCost(ctx context.Context, tenantID int64, modelName string, inputT
 	// Token / Tiered 计费
 	var inputCost, outputCost float64
 	if pricing.BillingMode == "tiered" {
-		inputCost, _ = calculateTieredCostFromPricing(ctx, modelName, inputTokens, true)
-		outputCost, _ = calculateTieredCostFromPricing(ctx, modelName, outputTokens, false)
+		if len(pricing.CustomTiers) > 0 {
+			inputCost = calculateTieredCostFromTiers(pricing.CustomTiers, inputTokens, true)
+			outputCost = calculateTieredCostFromTiers(pricing.CustomTiers, outputTokens, false)
+		} else {
+			inputCost, _ = calculateTieredCostFromPricing(ctx, modelName, inputTokens, true)
+			outputCost, _ = calculateTieredCostFromPricing(ctx, modelName, outputTokens, false)
+		}
 	} else {
 		// 统一单价模式
 		inputCost = float64(inputTokens) / 1_000_000.0 * pricing.InputPrice
@@ -326,7 +353,11 @@ func CalculateCostWithUsage(ctx context.Context, tenantID int64, modelName strin
 	// 基础输入费用
 	var baseInputCost float64
 	if pricing.BillingMode == "tiered" {
-		baseInputCost, _ = calculateTieredCostFromPricing(ctx, modelName, baseInputTokens, true)
+		if len(pricing.CustomTiers) > 0 {
+			baseInputCost = calculateTieredCostFromTiers(pricing.CustomTiers, baseInputTokens, true)
+		} else {
+			baseInputCost, _ = calculateTieredCostFromPricing(ctx, modelName, baseInputTokens, true)
+		}
 	} else {
 		baseInputCost = float64(baseInputTokens) / 1_000_000.0 * pricing.InputPrice
 	}
@@ -334,7 +365,11 @@ func CalculateCostWithUsage(ctx context.Context, tenantID int64, modelName strin
 	// 输出费用
 	var outputCost float64
 	if pricing.BillingMode == "tiered" {
-		outputCost, _ = calculateTieredCostFromPricing(ctx, modelName, outputTokens, false)
+		if len(pricing.CustomTiers) > 0 {
+			outputCost = calculateTieredCostFromTiers(pricing.CustomTiers, outputTokens, false)
+		} else {
+			outputCost, _ = calculateTieredCostFromPricing(ctx, modelName, outputTokens, false)
+		}
 	} else {
 		outputCost = float64(outputTokens) / 1_000_000.0 * pricing.OutputPrice
 	}
@@ -486,6 +521,15 @@ func calculateTieredCostFromPricing(ctx context.Context, modelName string, token
 		return 0, 1.0
 	}
 
+	return calculateTieredCostFromTiers(tiers, tokens, isInput), 1.0
+}
+
+// calculateTieredCostFromTiers 从给定的阶梯数组计算费用（租户自定义阶梯或基础阶梯共用）
+func calculateTieredCostFromTiers(tiers []pricingTierRow, tokens int, isInput bool) float64 {
+	if tokens <= 0 || len(tiers) == 0 {
+		return 0
+	}
+
 	var totalCost float64
 	remaining := int64(tokens)
 
@@ -500,7 +544,6 @@ func calculateTieredCostFromPricing(ctx context.Context, modelName string, token
 		}
 
 		if tier.MaxTokens == nil {
-			// 最后一段无上限
 			totalCost += float64(remaining) / 1_000_000.0 * price
 			remaining = 0
 		} else {
@@ -517,5 +560,5 @@ func calculateTieredCostFromPricing(ctx context.Context, modelName string, token
 		}
 	}
 
-	return totalCost, 1.0
+	return totalCost
 }

internal/logic/tenant/model.go

@@ -2,17 +2,27 @@ package tenant
 
 import (
 	"context"
+	"encoding/json"
 
 	v1 "github.com/qianfree/team-api/api/tenant/v1"
 	"github.com/qianfree/team-api/internal/dao"
 )
 
+// pricingTierRow 阶梯定价行（与 billing.pricingTierRow 结构一致）
+type pricingTierRow struct {
+	MinTokens   int64   `json:"min_tokens"`
+	MaxTokens   *int64  `json:"max_tokens"`
+	InputPrice  float64 `json:"input_price"`
+	OutputPrice float64 `json:"output_price"`
+}
+
 // ListAvailableModels 获取租户可用的模型列表
 func (s *sTenant) ListAvailableModels(ctx context.Context, req *v1.TenantAvailableModelsReq) (*v1.TenantAvailableModelsRes, error) {
 	tenantID := ctxTenantID(ctx)
 
 	query := dao.MdlTenantModels.Ctx(ctx).
 		LeftJoin("mdl_models m ON mdl_tenant_models.model_id = m.id").
+		LeftJoin("mdl_pricing p ON p.model_id = mdl_tenant_models.model_id AND p.min_tokens = 0").
 		Where("mdl_tenant_models.tenant_id", tenantID).
 		Where("mdl_tenant_models.enabled", true).
 		Where("m.status", "active")
@@ -25,23 +35,34 @@ func (s *sTenant) ListAvailableModels(ctx context.Context, req *v1.TenantAvailab
 	}
 
 	var results []struct {
-		ID               int64    `json:"id"`
-		ModelId          string   `json:"model_id"`
-		ModelName        string   `json:"model_name"`
-		Category         string   `json:"category"`
-		MaxContextTokens int      `json:"max_context_tokens"`
-		MaxOutputTokens  int      `json:"max_output_tokens"`
-		Description      string   `json:"description"`
-		Tags             string   `json:"tags"`
-		Capabilities     string   `json:"capabilities"`
-		BillingMode      *string  `json:"billing_mode"`
-		PerRequestPrice  *float64 `json:"per_request_price"`
-		DiscountRatio    *float64 `json:"discount_ratio"`
-		MaxConcurrency   *int     `json:"max_concurrency"`
+		ID                       int64    `json:"id"`
+		ModelDBID                int64    `json:"model_db_id"`
+		ModelId                  string   `json:"model_id"`
+		ModelName                string   `json:"model_name"`
+		Category                 string   `json:"category"`
+		MaxContextTokens         int      `json:"max_context_tokens"`
+		MaxOutputTokens          int      `json:"max_output_tokens"`
+		Description              string   `json:"description"`
+		Tags                     string   `json:"tags"`
+		Capabilities             string   `json:"capabilities"`
+		BillingMode              *string  `json:"billing_mode"`
+		PerRequestPrice          *float64 `json:"per_request_price"`
+		DiscountRatio            *float64 `json:"discount_ratio"`
+		MaxConcurrency           *int     `json:"max_concurrency"`
+		BaseInputPrice           float64  `json:"base_input_price"`
+		BaseOutputPrice          float64  `json:"base_output_price"`
+		BaseCacheReadPrice       float64  `json:"base_cache_read_price"`
+		BaseCacheCreationPrice   float64  `json:"base_cache_creation_price"`
+		BaseBillingMode          string   `json:"base_billing_mode"`
+		CustomInputPrice         *float64 `json:"custom_input_price"`
+		CustomOutputPrice        *float64 `json:"custom_output_price"`
+		CustomCacheReadPrice     *float64 `json:"custom_cache_read_price"`
+		CustomCacheCreationPrice *float64 `json:"custom_cache_creation_price"`
+		CustomPricingTiers       string   `json:"custom_pricing_tiers"`
 	}
 
 	err := query.
-		Fields("mdl_tenant_models.id, m.model_id, m.model_name, m.category, m.max_context_tokens, m.max_output_tokens, m.description, m.tags, m.capabilities, mdl_tenant_models.billing_mode, mdl_tenant_models.per_request_price, mdl_tenant_models.discount_ratio, mdl_tenant_models.max_concurrency").
+		Fields("mdl_tenant_models.id, mdl_tenant_models.model_id AS model_db_id, m.model_id, m.model_name, m.category, m.max_context_tokens, m.max_output_tokens, m.description, m.tags, m.capabilities, mdl_tenant_models.billing_mode, mdl_tenant_models.per_request_price, mdl_tenant_models.discount_ratio, mdl_tenant_models.max_concurrency, p.input_price AS base_input_price, p.output_price AS base_output_price, p.cache_read_price AS base_cache_read_price, p.cache_creation_price AS base_cache_creation_price, p.billing_mode AS base_billing_mode, mdl_tenant_models.custom_input_price, mdl_tenant_models.custom_output_price, mdl_tenant_models.custom_cache_read_price, mdl_tenant_models.custom_cache_creation_price, mdl_tenant_models.custom_pricing_tiers").
 		OrderAsc("m.category").
 		OrderAsc("m.model_id").
 		Scan(&results)
@@ -49,24 +70,141 @@ func (s *sTenant) ListAvailableModels(ctx context.Context, req *v1.TenantAvailab
 		return nil, err
 	}
 
+	// 收集需要查询 base 阶梯定价的 model 内部 ID（mdl_models.id）
+	tieredModelDBIDs := make([]int64, 0)
+	for _, r := range results {
+		effectiveBillingMode := resolveBillingMode(r.BillingMode, r.BaseBillingMode)
+		if effectiveBillingMode == "tiered" && r.CustomPricingTiers == "" {
+			tieredModelDBIDs = append(tieredModelDBIDs, r.ModelDBID)
+		}
+	}
+
+	// 批量查询 base 阶梯定价（min_tokens > 0 的阶梯）
+	baseTiersMap := make(map[int64][]v1.PricingTierItem)
+	if len(tieredModelDBIDs) > 0 {
+		var baseTiers []struct {
+			ModelId     int64   `json:"model_id"`
+			MinTokens   int64   `json:"min_tokens"`
+			MaxTokens   *int64  `json:"max_tokens"`
+			InputPrice  float64 `json:"input_price"`
+			OutputPrice float64 `json:"output_price"`
+		}
+		err = dao.MdlPricing.Ctx(ctx).
+			WhereIn("model_id", tieredModelDBIDs).
+			Where("billing_mode", "tiered").
+			Where("min_tokens > 0").
+			Fields("model_id, min_tokens, max_tokens, input_price, output_price").
+			OrderAsc("model_id").
+			OrderAsc("min_tokens").
+			Scan(&baseTiers)
+		if err == nil {
+			for _, t := range baseTiers {
+				baseTiersMap[t.ModelId] = append(baseTiersMap[t.ModelId], v1.PricingTierItem{
+					MinTokens:   t.MinTokens,
+					MaxTokens:   t.MaxTokens,
+					InputPrice:  t.InputPrice,
+					OutputPrice: t.OutputPrice,
+				})
+			}
+		}
+	}
+
 	list := make([]v1.TenantAvailableModelItem, 0, len(results))
 	for _, r := range results {
-		list = append(list, v1.TenantAvailableModelItem{
-			ID:              r.ID,
-			ModelId:         r.ModelId,
-			ModelName:       r.ModelName,
-			Category:        r.Category,
-			MaxContext:      r.MaxContextTokens,
-			MaxOutput:       r.MaxOutputTokens,
-			Description:     r.Description,
-			Tags:            r.Tags,
-			Capabilities:    r.Capabilities,
-			BillingMode:     r.BillingMode,
-			PerRequestPrice: r.PerRequestPrice,
-			DiscountRatio:   r.DiscountRatio,
-			MaxConcurrency:  r.MaxConcurrency,
-		})
+		effectiveBillingMode := resolveBillingMode(r.BillingMode, r.BaseBillingMode)
+
+		inputPrice := effectivePrice(r.CustomInputPrice, r.BaseInputPrice)
+		outputPrice := effectivePrice(r.CustomOutputPrice, r.BaseOutputPrice)
+		cacheReadPrice := effectivePrice(r.CustomCacheReadPrice, r.BaseCacheReadPrice)
+		cacheCreationPrice := effectivePrice(r.CustomCacheCreationPrice, r.BaseCacheCreationPrice)
+
+		item := v1.TenantAvailableModelItem{
+			ID:                 r.ID,
+			ModelId:            r.ModelId,
+			ModelName:          r.ModelName,
+			Category:           r.Category,
+			MaxContext:         r.MaxContextTokens,
+			MaxOutput:          r.MaxOutputTokens,
+			Description:        r.Description,
+			Tags:               r.Tags,
+			Capabilities:       r.Capabilities,
+			BillingMode:        &effectiveBillingMode,
+			PerRequestPrice:    r.PerRequestPrice,
+			DiscountRatio:      r.DiscountRatio,
+			MaxConcurrency:     r.MaxConcurrency,
+			InputPrice:         inputPrice,
+			OutputPrice:        outputPrice,
+			CacheReadPrice:     cacheReadPrice,
+			CacheCreationPrice: cacheCreationPrice,
+		}
+
+		// 阶梯定价处理
+		if effectiveBillingMode == "tiered" {
+			var tiers []v1.PricingTierItem
+
+			// 优先使用租户自定义阶梯
+			if r.CustomPricingTiers != "" && r.CustomPricingTiers != "null" && r.CustomPricingTiers != "[]" {
+				var raw []pricingTierRow
+				if json.Unmarshal([]byte(r.CustomPricingTiers), &raw) == nil && len(raw) > 0 {
+					for _, t := range raw {
+						tiers = append(tiers, v1.PricingTierItem{
+							MinTokens:   t.MinTokens,
+							MaxTokens:   t.MaxTokens,
+							InputPrice:  t.InputPrice,
+							OutputPrice: t.OutputPrice,
+						})
+					}
+				}
+			}
+
+			// 使用 base 阶梯定价
+			if len(tiers) == 0 {
+				// 第一阶梯（min_tokens=0，来自主查询的 base_input/output_price）
+				if r.BaseInputPrice > 0 || r.BaseOutputPrice > 0 {
+					tiers = append(tiers, v1.PricingTierItem{
+						MinTokens:   0,
+						MaxTokens:   nil,
+						InputPrice:  r.BaseInputPrice,
+						OutputPrice: r.BaseOutputPrice,
+					})
+				}
+				// 后续阶梯
+				if rest, ok := baseTiersMap[r.ModelDBID]; ok {
+					// 设置第一阶梯的 max_tokens
+					if len(tiers) > 0 && len(rest) > 0 {
+						tiers[0].MaxTokens = &rest[0].MinTokens
+					}
+					tiers = append(tiers, rest...)
+				}
+			}
+
+			item.PricingTiers = tiers
+		}
+
+		list = append(list, item)
 	}
 
 	return &v1.TenantAvailableModelsRes{List: list}, nil
 }
+
+// resolveBillingMode 解析有效计费模式
+func resolveBillingMode(tenantMode *string, baseMode string) string {
+	if tenantMode != nil && *tenantMode != "" {
+		return *tenantMode
+	}
+	if baseMode != "" {
+		return baseMode
+	}
+	return "token"
+}
+
+// effectivePrice 计算有效价格：自定义价优先，否则用基础价
+func effectivePrice(custom *float64, base float64) *float64 {
+	if custom != nil && *custom > 0 {
+		return custom
+	}
+	if base > 0 {
+		return &base
+	}
+	return nil
+}

internal/model/do/chn_channel_keys.go

@@ -20,7 +20,7 @@ type ChnChannelKeys struct {
 	LastUsedAt     *gtime.Time // 最后使用时间
 	LastError      any         // 最后一次错误信息
 	CreatedAt      *gtime.Time // 创建时间
-	UpdatedAt      *gtime.Time // 更新时间
 	KeyType        any         // Key 类型：apikey（传统静态密钥）/ oauth（OAuth 令牌）
 	TokenExpiresAt *gtime.Time // OAuth access_token 过期时间（仅 key_type=oauth 时有值）
+	UpdatedAt      *gtime.Time // 更新时间
 }

internal/model/do/mdl_models.go

@@ -21,10 +21,10 @@ type MdlModels struct {
 	MaxOutputTokens  any         // 最大输出 token 数
 	Description      any         // 模型描述
 	Tags             []string    // 标签（如 reasoning、vision、function_calling）
+	Capabilities     any         // 模型能力特性（如 vision、function_calling、reasoning 等）
 	CreatedAt        *gtime.Time // 创建时间
 	UpdatedAt        *gtime.Time // 更新时间
 	DeprecatedAt     *gtime.Time // 标记弃用的时间（NULL表示未弃用）
 	SunsetDate       *gtime.Time // 计划下线日期（到达后返回410 Gone，NULL表示未设置）
 	ReplacementModel any         // 推荐替代模型名（NULL表示无替代）
-	Capabilities     any         // 模型能力特性（如 vision、function_calling、reasoning 等）
 }

internal/model/do/mdl_tenant_models.go

@@ -11,19 +11,22 @@ import (
 
 // MdlTenantModels is the golang structure of table mdl_tenant_models for DAO operations like Where/Data.
 type MdlTenantModels struct {
-	g.Meta            `orm:"table:mdl_tenant_models, do:true"`
-	Id                any         // 主键ID
-	TenantId          any         // 租户ID
-	ModelId           any         // 模型ID
-	Enabled           any         // 是否启用（禁用后该租户无法调用此模型）
-	CustomInputPrice  any         // 租户自定义输入价格（NULL 表示使用默认定价）
-	CustomOutputPrice any         // 租户自定义输出价格（NULL 表示使用默认定价）
-	Multiplier        any         // 租户价格倍率（VIP 折扣等，最终价格 = 基础价格 × 模型倍率 × 租户倍率）
-	CreatedAt         *gtime.Time // 创建时间
-	UpdatedAt         *gtime.Time // 更新时间
-	BillingMode       any         // 覆盖模型计费方式（NULL表示跟随模型默认）
-	PerRequestPrice   any         // 按次计费单价（覆盖模型默认，仅 billing_mode = per_request 时有效）
-	DiscountRatio     any         // 折扣比例（如0.8表示八折，NULL表示不打折，优先于 multiplier 使用）
-	MaxConcurrency    any         // 单模型并发上限（NULL表示不限制）
-	ChannelScope      any         // 渠道范围覆盖（NULL跟随租户默认，[]表示全部，数组表示指定渠道ID）
+	g.Meta                   `orm:"table:mdl_tenant_models, do:true"`
+	Id                       any         // 主键ID
+	TenantId                 any         // 租户ID
+	ModelId                  any         // 模型ID
+	Enabled                  any         // 是否启用（禁用后该租户无法调用此模型）
+	CustomInputPrice         any         // 租户自定义输入价格（NULL 表示使用默认定价）
+	CustomOutputPrice        any         // 租户自定义输出价格（NULL 表示使用默认定价）
+	Multiplier               any         // 租户价格倍率（VIP 折扣等，最终价格 = 基础价格 × 模型倍率 × 租户倍率）
+	CreatedAt                *gtime.Time // 创建时间
+	UpdatedAt                *gtime.Time // 更新时间
+	BillingMode              any         // 覆盖模型计费方式（NULL表示跟随模型默认）
+	PerRequestPrice          any         // 按次计费单价（覆盖模型默认，仅 billing_mode = per_request 时有效）
+	DiscountRatio            any         // 折扣比例（如0.8表示八折，NULL表示不打折，优先于 multiplier 使用）
+	MaxConcurrency           any         // 单模型并发上限（NULL表示不限制）
+	ChannelScope             any         // 渠道范围覆盖（NULL跟随租户默认，[]表示全部，数组表示指定渠道ID）
+	CustomCacheReadPrice     any         // 自定义缓存读取价格（$/1M token），NULL 表示使用基础定价
+	CustomCacheCreationPrice any         // 自定义缓存创建价格（$/1M token），NULL 表示使用基础定价
+	CustomPricingTiers       any         // 自定义阶梯定价（JSONB 数组），格式: [{"min_tokens":0,"max_tokens":100000,"input_price":0.5,"output_price":1.5,"cache_read_price":0.1,"cache_creation_price":0.2}]
 }

internal/model/do/ntf_messages.go

@@ -22,6 +22,6 @@ type NtfMessages struct {
 	IsRead      any         // 是否已读：0=未读, 1=已读
 	IsBroadcast any         // 是否广播消息：0=个人消息, 1=广播消息
 	Metadata    any         // 附加元数据（JSONB，如关联资源ID、跳转链接等）
-	CreatedAt   *gtime.Time // 创建时间
 	TargetRoles any         // 目标角色（NULL=全部角色，逗号分隔如 owner,admin 表示仅限这些角色）
+	CreatedAt   *gtime.Time // 创建时间
 }

internal/model/do/sys_plugins.go

@@ -0,0 +1,25 @@
+// =================================================================================
+// Code generated and maintained by GoFrame CLI tool. DO NOT EDIT.
+// =================================================================================
+
+package do
+
+import (
+	"github.com/gogf/gf/v2/frame/g"
+	"github.com/gogf/gf/v2/os/gtime"
+)
+
+// SysPlugins is the golang structure of table sys_plugins for DAO operations like Where/Data.
+type SysPlugins struct {
+	g.Meta    `orm:"table:sys_plugins, do:true"`
+	Id        any         //
+	Name      any         // 插件唯一标识，如 email-report
+	Label     any         // 显示名称
+	Version   any         // 当前安装版本
+	Status    any         // 状态：registered=已注册, installed=已安装, enabled=已启用, disabled=已禁用, error=异常
+	Category  any         // 分类：relay=代理扩展, middleware=中间件, billing=计费, notification=通知, extension=通用扩展
+	Config    any         // 插件全局配置（JSON）
+	ErrorMsg  any         // 异常信息
+	CreatedAt *gtime.Time //
+	UpdatedAt *gtime.Time //
+}

internal/model/do/tnt_tenant_plugins.go

@@ -0,0 +1,22 @@
+// =================================================================================
+// Code generated and maintained by GoFrame CLI tool. DO NOT EDIT.
+// =================================================================================
+
+package do
+
+import (
+	"github.com/gogf/gf/v2/frame/g"
+	"github.com/gogf/gf/v2/os/gtime"
+)
+
+// TntTenantPlugins is the golang structure of table tnt_tenant_plugins for DAO operations like Where/Data.
+type TntTenantPlugins struct {
+	g.Meta     `orm:"table:tnt_tenant_plugins, do:true"`
+	Id         any         //
+	TenantId   any         //
+	PluginName any         // 插件标识
+	Enabled    any         // 是否启用
+	Config     any         // 租户级配置覆盖（JSON），优先级高于全局配置
+	CreatedAt  *gtime.Time //
+	UpdatedAt  *gtime.Time //
+}

internal/model/entity/chn_channel_keys.go

@@ -18,7 +18,7 @@ type ChnChannelKeys struct {
 	LastUsedAt     *gtime.Time `json:"last_used_at"     orm:"last_used_at"     description:"最后使用时间"`                                        // 最后使用时间
 	LastError      string      `json:"last_error"       orm:"last_error"       description:"最后一次错误信息"`                                      // 最后一次错误信息
 	CreatedAt      *gtime.Time `json:"created_at"       orm:"created_at"       description:"创建时间"`                                          // 创建时间
-	UpdatedAt      *gtime.Time `json:"updated_at"       orm:"updated_at"       description:"更新时间"`                                          // 更新时间
 	KeyType        string      `json:"key_type"         orm:"key_type"         description:"Key 类型：apikey（传统静态密钥）/ oauth（OAuth 令牌）"`        // Key 类型：apikey（传统静态密钥）/ oauth（OAuth 令牌）
 	TokenExpiresAt *gtime.Time `json:"token_expires_at" orm:"token_expires_at" description:"OAuth access_token 过期时间（仅 key_type=oauth 时有值）"` // OAuth access_token 过期时间（仅 key_type=oauth 时有值）
+	UpdatedAt      *gtime.Time `json:"updated_at"       orm:"updated_at"       description:"更新时间"`                                          // 更新时间
 }

internal/model/entity/mdl_models.go

@@ -19,10 +19,10 @@ type MdlModels struct {
 	MaxOutputTokens  int         `json:"max_output_tokens"  orm:"max_output_tokens"  description:"最大输出 token 数"`                                                    // 最大输出 token 数
 	Description      string      `json:"description"        orm:"description"        description:"模型描述"`                                                            // 模型描述
 	Tags             []string    `json:"tags"               orm:"tags"               description:"标签（如 reasoning、vision、function_calling）"`                         // 标签（如 reasoning、vision、function_calling）
+	Capabilities     string      `json:"capabilities"       orm:"capabilities"       description:"模型能力特性（如 vision、function_calling、reasoning 等）"`                   // 模型能力特性（如 vision、function_calling、reasoning 等）
 	CreatedAt        *gtime.Time `json:"created_at"         orm:"created_at"         description:"创建时间"`                                                            // 创建时间
 	UpdatedAt        *gtime.Time `json:"updated_at"         orm:"updated_at"         description:"更新时间"`                                                            // 更新时间
 	DeprecatedAt     *gtime.Time `json:"deprecated_at"      orm:"deprecated_at"      description:"标记弃用的时间（NULL表示未弃用）"`                                              // 标记弃用的时间（NULL表示未弃用）
 	SunsetDate       *gtime.Time `json:"sunset_date"        orm:"sunset_date"        description:"计划下线日期（到达后返回410 Gone，NULL表示未设置）"`                                 // 计划下线日期（到达后返回410 Gone，NULL表示未设置）
 	ReplacementModel string      `json:"replacement_model"  orm:"replacement_model"  description:"推荐替代模型名（NULL表示无替代）"`                                              // 推荐替代模型名（NULL表示无替代）
-	Capabilities     string      `json:"capabilities"       orm:"capabilities"       description:"模型能力特性（如 vision、function_calling、reasoning 等）"`                   // 模型能力特性（如 vision、function_calling、reasoning 等）
 }

internal/model/entity/mdl_tenant_models.go

@@ -10,18 +10,21 @@ import (
 
 // MdlTenantModels is the golang structure for table mdl_tenant_models.
 type MdlTenantModels struct {
-	Id                int64       `json:"id"                  orm:"id"                  description:"主键ID"`                                            // 主键ID
-	TenantId          int64       `json:"tenant_id"           orm:"tenant_id"           description:"租户ID"`                                            // 租户ID
-	ModelId           int64       `json:"model_id"            orm:"model_id"            description:"模型ID"`                                            // 模型ID
-	Enabled           bool        `json:"enabled"             orm:"enabled"             description:"是否启用（禁用后该租户无法调用此模型）"`                             // 是否启用（禁用后该租户无法调用此模型）
-	CustomInputPrice  float64     `json:"custom_input_price"  orm:"custom_input_price"  description:"租户自定义输入价格（NULL 表示使用默认定价）"`                        // 租户自定义输入价格（NULL 表示使用默认定价）
-	CustomOutputPrice float64     `json:"custom_output_price" orm:"custom_output_price" description:"租户自定义输出价格（NULL 表示使用默认定价）"`                        // 租户自定义输出价格（NULL 表示使用默认定价）
-	Multiplier        float64     `json:"multiplier"          orm:"multiplier"          description:"租户价格倍率（VIP 折扣等，最终价格 = 基础价格 × 模型倍率 × 租户倍率）"`       // 租户价格倍率（VIP 折扣等，最终价格 = 基础价格 × 模型倍率 × 租户倍率）
-	CreatedAt         *gtime.Time `json:"created_at"          orm:"created_at"          description:"创建时间"`                                            // 创建时间
-	UpdatedAt         *gtime.Time `json:"updated_at"          orm:"updated_at"          description:"更新时间"`                                            // 更新时间
-	BillingMode       string      `json:"billing_mode"        orm:"billing_mode"        description:"覆盖模型计费方式（NULL表示跟随模型默认）"`                          // 覆盖模型计费方式（NULL表示跟随模型默认）
-	PerRequestPrice   float64     `json:"per_request_price"   orm:"per_request_price"   description:"按次计费单价（覆盖模型默认，仅 billing_mode = per_request 时有效）"` // 按次计费单价（覆盖模型默认，仅 billing_mode = per_request 时有效）
-	DiscountRatio     float64     `json:"discount_ratio"      orm:"discount_ratio"      description:"折扣比例（如0.8表示八折，NULL表示不打折，优先于 multiplier 使用）"`      // 折扣比例（如0.8表示八折，NULL表示不打折，优先于 multiplier 使用）
-	MaxConcurrency    int         `json:"max_concurrency"     orm:"max_concurrency"     description:"单模型并发上限（NULL表示不限制）"`                              // 单模型并发上限（NULL表示不限制）
-	ChannelScope      string      `json:"channel_scope"       orm:"channel_scope"       description:"渠道范围覆盖（NULL跟随租户默认，[]表示全部，数组表示指定渠道ID）"`            // 渠道范围覆盖（NULL跟随租户默认，[]表示全部，数组表示指定渠道ID）
+	Id                       int64       `json:"id"                          orm:"id"                          description:"主键ID"`                                                                                                                                                            // 主键ID
+	TenantId                 int64       `json:"tenant_id"                   orm:"tenant_id"                   description:"租户ID"`                                                                                                                                                            // 租户ID
+	ModelId                  int64       `json:"model_id"                    orm:"model_id"                    description:"模型ID"`                                                                                                                                                            // 模型ID
+	Enabled                  bool        `json:"enabled"                     orm:"enabled"                     description:"是否启用（禁用后该租户无法调用此模型）"`                                                                                                                                             // 是否启用（禁用后该租户无法调用此模型）
+	CustomInputPrice         float64     `json:"custom_input_price"          orm:"custom_input_price"          description:"租户自定义输入价格（NULL 表示使用默认定价）"`                                                                                                                                        // 租户自定义输入价格（NULL 表示使用默认定价）
+	CustomOutputPrice        float64     `json:"custom_output_price"         orm:"custom_output_price"         description:"租户自定义输出价格（NULL 表示使用默认定价）"`                                                                                                                                        // 租户自定义输出价格（NULL 表示使用默认定价）
+	Multiplier               float64     `json:"multiplier"                  orm:"multiplier"                  description:"租户价格倍率（VIP 折扣等，最终价格 = 基础价格 × 模型倍率 × 租户倍率）"`                                                                                                                       // 租户价格倍率（VIP 折扣等，最终价格 = 基础价格 × 模型倍率 × 租户倍率）
+	CreatedAt                *gtime.Time `json:"created_at"                  orm:"created_at"                  description:"创建时间"`                                                                                                                                                            // 创建时间
+	UpdatedAt                *gtime.Time `json:"updated_at"                  orm:"updated_at"                  description:"更新时间"`                                                                                                                                                            // 更新时间
+	BillingMode              string      `json:"billing_mode"                orm:"billing_mode"                description:"覆盖模型计费方式（NULL表示跟随模型默认）"`                                                                                                                                          // 覆盖模型计费方式（NULL表示跟随模型默认）
+	PerRequestPrice          float64     `json:"per_request_price"           orm:"per_request_price"           description:"按次计费单价（覆盖模型默认，仅 billing_mode = per_request 时有效）"`                                                                                                                 // 按次计费单价（覆盖模型默认，仅 billing_mode = per_request 时有效）
+	DiscountRatio            float64     `json:"discount_ratio"              orm:"discount_ratio"              description:"折扣比例（如0.8表示八折，NULL表示不打折，优先于 multiplier 使用）"`                                                                                                                      // 折扣比例（如0.8表示八折，NULL表示不打折，优先于 multiplier 使用）
+	MaxConcurrency           int         `json:"max_concurrency"             orm:"max_concurrency"             description:"单模型并发上限（NULL表示不限制）"`                                                                                                                                              // 单模型并发上限（NULL表示不限制）
+	ChannelScope             string      `json:"channel_scope"               orm:"channel_scope"               description:"渠道范围覆盖（NULL跟随租户默认，[]表示全部，数组表示指定渠道ID）"`                                                                                                                            // 渠道范围覆盖（NULL跟随租户默认，[]表示全部，数组表示指定渠道ID）
+	CustomCacheReadPrice     float64     `json:"custom_cache_read_price"     orm:"custom_cache_read_price"     description:"自定义缓存读取价格（$/1M token），NULL 表示使用基础定价"`                                                                                                                             // 自定义缓存读取价格（$/1M token），NULL 表示使用基础定价
+	CustomCacheCreationPrice float64     `json:"custom_cache_creation_price" orm:"custom_cache_creation_price" description:"自定义缓存创建价格（$/1M token），NULL 表示使用基础定价"`                                                                                                                             // 自定义缓存创建价格（$/1M token），NULL 表示使用基础定价
+	CustomPricingTiers       string      `json:"custom_pricing_tiers"        orm:"custom_pricing_tiers"        description:"自定义阶梯定价（JSONB 数组），格式: [{\"min_tokens\":0,\"max_tokens\":100000,\"input_price\":0.5,\"output_price\":1.5,\"cache_read_price\":0.1,\"cache_creation_price\":0.2}]"` // 自定义阶梯定价（JSONB 数组），格式: [{"min_tokens":0,"max_tokens":100000,"input_price":0.5,"output_price":1.5,"cache_read_price":0.1,"cache_creation_price":0.2}]
 }

internal/model/entity/ntf_messages.go

@@ -20,6 +20,6 @@ type NtfMessages struct {
 	IsRead      int         `json:"is_read"      orm:"is_read"      description:"是否已读：0=未读, 1=已读"`                                                              // 是否已读：0=未读, 1=已读
 	IsBroadcast int         `json:"is_broadcast" orm:"is_broadcast" description:"是否广播消息：0=个人消息, 1=广播消息"`                                                        // 是否广播消息：0=个人消息, 1=广播消息
 	Metadata    string      `json:"metadata"     orm:"metadata"     description:"附加元数据（JSONB，如关联资源ID、跳转链接等）"`                                                   // 附加元数据（JSONB，如关联资源ID、跳转链接等）
-	CreatedAt   *gtime.Time `json:"created_at"   orm:"created_at"   description:"创建时间"`                                                                         // 创建时间
 	TargetRoles string      `json:"target_roles" orm:"target_roles" description:"目标角色（NULL=全部角色，逗号分隔如 owner,admin 表示仅限这些角色）"`                                   // 目标角色（NULL=全部角色，逗号分隔如 owner,admin 表示仅限这些角色）
+	CreatedAt   *gtime.Time `json:"created_at"   orm:"created_at"   description:"创建时间"`                                                                         // 创建时间
 }

internal/model/entity/sys_plugins.go

@@ -0,0 +1,23 @@
+// =================================================================================
+// Code generated and maintained by GoFrame CLI tool. DO NOT EDIT.
+// =================================================================================
+
+package entity
+
+import (
+	"github.com/gogf/gf/v2/os/gtime"
+)
+
+// SysPlugins is the golang structure for table sys_plugins.
+type SysPlugins struct {
+	Id        int64       `json:"id"         orm:"id"         description:""`                                                                           //
+	Name      string      `json:"name"       orm:"name"       description:"插件唯一标识，如 email-report"`                                                      // 插件唯一标识，如 email-report
+	Label     string      `json:"label"      orm:"label"      description:"显示名称"`                                                                       // 显示名称
+	Version   string      `json:"version"    orm:"version"    description:"当前安装版本"`                                                                     // 当前安装版本
+	Status    string      `json:"status"     orm:"status"     description:"状态：registered=已注册, installed=已安装, enabled=已启用, disabled=已禁用, error=异常"`      // 状态：registered=已注册, installed=已安装, enabled=已启用, disabled=已禁用, error=异常
+	Category  string      `json:"category"   orm:"category"   description:"分类：relay=代理扩展, middleware=中间件, billing=计费, notification=通知, extension=通用扩展"` // 分类：relay=代理扩展, middleware=中间件, billing=计费, notification=通知, extension=通用扩展
+	Config    string      `json:"config"     orm:"config"     description:"插件全局配置（JSON）"`                                                               // 插件全局配置（JSON）
+	ErrorMsg  string      `json:"error_msg"  orm:"error_msg"  description:"异常信息"`                                                                       // 异常信息
+	CreatedAt *gtime.Time `json:"created_at" orm:"created_at" description:""`                                                                           //
+	UpdatedAt *gtime.Time `json:"updated_at" orm:"updated_at" description:""`                                                                           //
+}

internal/model/entity/tnt_tenant_plugins.go

@@ -0,0 +1,20 @@
+// =================================================================================
+// Code generated and maintained by GoFrame CLI tool. DO NOT EDIT.
+// =================================================================================
+
+package entity
+
+import (
+	"github.com/gogf/gf/v2/os/gtime"
+)
+
+// TntTenantPlugins is the golang structure for table tnt_tenant_plugins.
+type TntTenantPlugins struct {
+	Id         int64       `json:"id"          orm:"id"          description:""`                        //
+	TenantId   int64       `json:"tenant_id"   orm:"tenant_id"   description:""`                        //
+	PluginName string      `json:"plugin_name" orm:"plugin_name" description:"插件标识"`                    // 插件标识
+	Enabled    bool        `json:"enabled"     orm:"enabled"     description:"是否启用"`                    // 是否启用
+	Config     string      `json:"config"      orm:"config"      description:"租户级配置覆盖（JSON），优先级高于全局配置"` // 租户级配置覆盖（JSON），优先级高于全局配置
+	CreatedAt  *gtime.Time `json:"created_at"  orm:"created_at"  description:""`                        //
+	UpdatedAt  *gtime.Time `json:"updated_at"  orm:"updated_at"  description:""`                        //
+}

migrations/000002_add_tenant_model_custom_pricing.sql

@@ -0,0 +1,22 @@
+-- +goose Up
+-- 租户模型分配增加自定义缓存定价覆盖字段和阶梯定价支持
+
+-- 缓存定价覆盖
+ALTER TABLE mdl_tenant_models
+    ADD COLUMN IF NOT EXISTS custom_cache_read_price     NUMERIC(20,10),
+    ADD COLUMN IF NOT EXISTS custom_cache_creation_price  NUMERIC(20,10);
+
+COMMENT ON COLUMN mdl_tenant_models.custom_cache_read_price IS '自定义缓存读取价格（$/1M token），NULL 表示使用基础定价';
+COMMENT ON COLUMN mdl_tenant_models.custom_cache_creation_price IS '自定义缓存创建价格（$/1M token），NULL 表示使用基础定价';
+
+-- 阶梯定价
+ALTER TABLE mdl_tenant_models
+    ADD COLUMN IF NOT EXISTS custom_pricing_tiers JSONB;
+
+COMMENT ON COLUMN mdl_tenant_models.custom_pricing_tiers IS '自定义阶梯定价（JSONB 数组），格式: [{"min_tokens":0,"max_tokens":100000,"input_price":0.5,"output_price":1.5,"cache_read_price":0.1,"cache_creation_price":0.2}]';
+
+-- +goose Down
+ALTER TABLE mdl_tenant_models
+    DROP COLUMN IF EXISTS custom_cache_read_price,
+    DROP COLUMN IF EXISTS custom_cache_creation_price,
+    DROP COLUMN IF EXISTS custom_pricing_tiers;

web/admin/src/views/tenant/TenantDetailPage.vue

@@ -1,5 +1,5 @@
 <script setup lang="ts">
-import { ref, reactive, computed, onMounted, h } from 'vue'
+import { ref, reactive, computed, onMounted, h, watch } from 'vue'
 import { useRoute, useRouter } from 'vue-router'
 import { Tag, Button, Space, Popconfirm, Message } from '@arco-design/web-vue'
 import type { TableColumnData } from '@arco-design/web-vue'
@@ -199,9 +199,15 @@ const editModelForm = reactive({
   max_concurrency: 5 as number | null,
   custom_input_price: null as number | null,
   custom_output_price: null as number | null,
+  custom_cache_read_price: null as number | null,
+  custom_cache_creation_price: null as number | null,
+  custom_pricing_tiers: [] as any[],
 })
 
+let suppressBillingWatch = false
+
 function openEditModel(record: any) {
+  suppressBillingWatch = true
   editingModel.value = record
   editModelForm.enabled = record.enabled
   editModelForm.billing_mode = record.billing_mode || null
@@ -210,10 +216,48 @@ function openEditModel(record: any) {
   editModelForm.max_concurrency = record.max_concurrency ?? 5
   editModelForm.custom_input_price = record.custom_input_price || null
   editModelForm.custom_output_price = record.custom_output_price || null
+  editModelForm.custom_cache_read_price = record.custom_cache_read_price || null
+  editModelForm.custom_cache_creation_price = record.custom_cache_creation_price || null
+  editModelForm.custom_pricing_tiers = record.custom_pricing_tiers?.length > 0 ? [...record.custom_pricing_tiers] : []
   showEditModelModal.value = true
+  suppressBillingWatch = false
 }
 
-async function handleEditModel(done: () => void) {
+function addTenantTier() {
+  const tiers = editModelForm.custom_pricing_tiers
+  const last = tiers[tiers.length - 1]
+  const newMin = last?.max_tokens ?? 0
+  tiers.push({
+    min_tokens: newMin,
+    max_tokens: null,
+    input_price: 0,
+    output_price: 0,
+    cache_read_price: 0,
+    cache_creation_price: 0,
+  })
+  if (last && last.max_tokens === null) {
+    last.max_tokens = newMin
+  }
+}
+
+// 计费模式切换时清理互斥字段
+watch(() => editModelForm.billing_mode, (mode) => {
+  if (suppressBillingWatch) return
+  if (!mode) {
+    // 默认模式：清空自定义价格，保留折扣
+    editModelForm.custom_input_price = null
+    editModelForm.custom_output_price = null
+    editModelForm.custom_cache_read_price = null
+    editModelForm.custom_cache_creation_price = null
+    editModelForm.per_request_price = null
+    editModelForm.custom_pricing_tiers = []
+  } else {
+    // 自定义价格模式：清空折扣比例
+    editModelForm.discount_ratio = null
+  }
+})
+
+async function handleEditModel() {
   if (!editingModel.value) return
   editModelLoading.value = true
   try {
@@ -223,10 +267,10 @@ async function handleEditModel(done: () => void) {
     }
     await request.put(`/admin/tenants/${tenantId}/models/${editingModel.value.model_id}`, payload)
     Message.success('更新成功')
-    done()
+    showEditModelModal.value = false
     fetchTenantModels()
   } catch {
-    return false
+    // error handled by interceptor
   } finally {
     editModelLoading.value = false
   }
@@ -548,49 +592,132 @@ onMounted(() => {
       />
     </AModal>
 
-    <!-- Edit Model Config Modal -->
-    <AModal
+    <!-- Edit Model Config Drawer -->
+    <ADrawer
       v-model:visible="showEditModelModal"
       :title="`编辑模型 - ${editingModel?.model_name || ''}`"
-      :width="550"
-      :on-before-ok="handleEditModel"
-      :ok-loading="editModelLoading"
+      :width="640"
+      :mask-closable="false"
+      :footer="true"
     >
       <AForm :model="editModelForm" :auto-label-width="true" layout="vertical">
         <AFormItem label="启用">
           <ASwitch v-model="editModelForm.enabled" />
         </AFormItem>
-        <AFormItem label="计费模式">
-          <ASelect
-            v-model="editModelForm.billing_mode"
-            :options="[{ label: '默认（跟随模型）', value: '' }, { label: 'Token 计费', value: 'token' }, { label: '按次计费', value: 'per_request' }]"
-            allow-clear
-            placeholder="默认"
-          />
-        </AFormItem>
-        <AFormItem label="折扣比例">
-          <AInputNumber v-model="editModelForm.discount_ratio" :min="0" :max="1" :step="0.05" :precision="2" placeholder="如 0.8 = 8折" class="w-full" />
-        </AFormItem>
-        <AFormItem label="自定义输入价格">
-          <AInputNumber v-model="editModelForm.custom_input_price" :min="0" :precision="4" placeholder="留空则使用默认价格" class="w-full">
-            <template #suffix>$ / 1M</template>
-          </AInputNumber>
-        </AFormItem>
-        <AFormItem label="自定义输出价格">
-          <AInputNumber v-model="editModelForm.custom_output_price" :min="0" :precision="4" placeholder="留空则使用默认价格" class="w-full">
-            <template #suffix>$ / 1M</template>
-          </AInputNumber>
-        </AFormItem>
-        <AFormItem label="按次单价">
-          <AInputNumber v-model="editModelForm.per_request_price" :min="0" :precision="4" placeholder="仅按次计费模式有效" class="w-full">
-            <template #suffix>$ / 次</template>
-          </AInputNumber>
-        </AFormItem>
         <AFormItem label="单模型并发上限">
           <AInputNumber v-model="editModelForm.max_concurrency" :min="0" placeholder="默认5" class="w-full" />
         </AFormItem>
+
+        <AFormItem label="计费模式">
+          <ARadioGroup v-model="editModelForm.billing_mode" type="button">
+            <ARadio value="">默认</ARadio>
+            <ARadio value="token">Token</ARadio>
+            <ARadio value="per_request">按次</ARadio>
+            <ARadio value="tiered">阶梯</ARadio>
+          </ARadioGroup>
+        </AFormItem>
+
+        <!-- 默认模式：仅折扣比例 -->
+        <template v-if="!editModelForm.billing_mode">
+          <ADivider margin="8px">折扣</ADivider>
+          <div style="color: var(--ta-text-tertiary); font-size: 12px; margin-bottom: 12px">使用模型基础定价 × 折扣比例</div>
+          <AFormItem label="折扣比例">
+            <AInputNumber v-model="editModelForm.discount_ratio" :min="0" :max="1" :step="0.05" :precision="2" placeholder="如 0.8 = 8折" class="w-full" />
+          </AFormItem>
+        </template>
+
+        <!-- Token 模式：自定义价格 -->
+        <template v-if="editModelForm.billing_mode === 'token'">
+          <ADivider margin="8px">自定义价格</ADivider>
+          <div style="color: var(--ta-text-tertiary); font-size: 12px; margin-bottom: 12px">留空则使用模型基础定价</div>
+          <div class="grid grid-cols-2 gap-x-3">
+            <AFormItem label="输入价格">
+              <AInputNumber v-model="editModelForm.custom_input_price" :min="0" :precision="4" placeholder="默认" class="w-full">
+                <template #suffix>$ / 1M</template>
+              </AInputNumber>
+            </AFormItem>
+            <AFormItem label="输出价格">
+              <AInputNumber v-model="editModelForm.custom_output_price" :min="0" :precision="4" placeholder="默认" class="w-full">
+                <template #suffix>$ / 1M</template>
+              </AInputNumber>
+            </AFormItem>
+            <AFormItem label="缓存读取价格">
+              <AInputNumber v-model="editModelForm.custom_cache_read_price" :min="0" :precision="4" placeholder="默认" class="w-full">
+                <template #suffix>$ / 1M</template>
+              </AInputNumber>
+            </AFormItem>
+            <AFormItem label="缓存创建价格">
+              <AInputNumber v-model="editModelForm.custom_cache_creation_price" :min="0" :precision="4" placeholder="默认" class="w-full">
+                <template #suffix>$ / 1M</template>
+              </AInputNumber>
+            </AFormItem>
+          </div>
+        </template>
+
+        <!-- 按次计费 -->
+        <template v-if="editModelForm.billing_mode === 'per_request'">
+          <ADivider margin="8px">按次定价</ADivider>
+          <AFormItem label="按次单价">
+            <AInputNumber v-model="editModelForm.per_request_price" :min="0" :precision="4" placeholder="每次调用价格" class="w-full">
+              <template #suffix>$ / 次</template>
+            </AInputNumber>
+          </AFormItem>
+        </template>
+
+        <!-- 阶梯计费 -->
+        <template v-if="editModelForm.billing_mode === 'tiered'">
+          <ADivider margin="8px">阶梯定价</ADivider>
+          <div style="color: var(--ta-text-tertiary); font-size: 12px; margin-bottom: 12px">按 Token 用量分段设置不同价格，留空则使用模型基础阶梯定价</div>
+          <div v-for="(tier, index) in editModelForm.custom_pricing_tiers" :key="index" class="tier-card">
+            <div class="tier-header">
+              <span class="tier-label">第 {{ index + 1 }} 梯</span>
+              <AButton v-if="editModelForm.custom_pricing_tiers.length > 1" size="mini" status="danger" @click="editModelForm.custom_pricing_tiers.splice(index, 1)">删除</AButton>
+            </div>
+            <div class="grid grid-cols-2 gap-x-3">
+              <AFormItem label="起始 Token">
+                <AInputNumber v-model="tier.min_tokens" :min="0" :step="1000" placeholder="0" class="w-full" />
+              </AFormItem>
+              <AFormItem label="结束 Token">
+                <AInputNumber v-if="index < editModelForm.custom_pricing_tiers.length - 1" v-model="tier.max_tokens" :min="0" :step="1000" placeholder="上限" class="w-full" />
+                <AInput v-else model-value="无上限" disabled class="w-full" />
+              </AFormItem>
+            </div>
+            <div class="grid grid-cols-2 gap-x-3 mt-2">
+              <AFormItem label="输入价格">
+                <AInputNumber v-model="tier.input_price" :min="0" :precision="4" class="w-full">
+                  <template #suffix>$/1M</template>
+                </AInputNumber>
+              </AFormItem>
+              <AFormItem label="输出价格">
+                <AInputNumber v-model="tier.output_price" :min="0" :precision="4" class="w-full">
+                  <template #suffix>$/1M</template>
+                </AInputNumber>
+              </AFormItem>
+            </div>
+            <div class="grid grid-cols-2 gap-x-3 mt-2">
+              <AFormItem label="缓存读取价格">
+                <AInputNumber v-model="tier.cache_read_price" :min="0" :precision="4" class="w-full">
+                  <template #suffix>$/1M</template>
+                </AInputNumber>
+              </AFormItem>
+              <AFormItem label="缓存创建价格">
+                <AInputNumber v-model="tier.cache_creation_price" :min="0" :precision="4" class="w-full">
+                  <template #suffix>$/1M</template>
+                </AInputNumber>
+              </AFormItem>
+            </div>
+          </div>
+          <AButton type="dashed" long @click="addTenantTier" class="mt-2">+ 添加梯度</AButton>
+        </template>
+
       </AForm>
-    </AModal>
+      <template #footer>
+        <ASpace>
+          <AButton @click="showEditModelModal = false">取消</AButton>
+          <AButton type="primary" :loading="editModelLoading" @click="handleEditModel">保存</AButton>
+        </ASpace>
+      </template>
+    </ADrawer>
 
     <!-- Recharge / Adjust Balance Modal -->
     <AModal v-model:visible="showRechargeModal" title="充值 / 调整余额" :width="420" :on-before-ok="handleRecharge" :ok-loading="rechargeLoading">
@@ -640,4 +767,21 @@ onMounted(() => {
   padding-top: 16px;
   border-top: 1px solid var(--ta-border-light);
 }
+.tier-card {
+  padding: 12px 16px;
+  background: var(--color-fill-1);
+  border-radius: 8px;
+  margin-bottom: 8px;
+}
+.tier-header {
+  display: flex;
+  justify-content: space-between;
+  align-items: center;
+  margin-bottom: 8px;
+}
+.tier-label {
+  font-size: 13px;
+  font-weight: 600;
+  color: var(--ta-text-secondary);
+}
 </style>

web/tenant/src/views/service/ModelsPage.vue

@@ -3,6 +3,13 @@ import { ref, computed, onMounted } from 'vue'
 import Icon from '@/components/common/Icon.vue'
 import request from '@/utils/request'
 
+interface PricingTierItem {
+	min_tokens: number
+	max_tokens: number | null
+	input_price: number
+	output_price: number
+}
+
 interface ModelItem {
 	id: number
 	model_id: string
@@ -17,6 +24,11 @@ interface ModelItem {
 	per_request_price: number | null
 	discount_ratio: number | null
 	max_concurrency: number | null
+	input_price: number | null
+	output_price: number | null
+	cache_read_price: number | null
+	cache_creation_price: number | null
+	pricing_tiers: PricingTierItem[]
 }
 
 const capabilityLabel: Record<string, string> = {
@@ -71,6 +83,11 @@ const models = ref<ModelItem[]>([])
 const loading = ref(false)
 const searchQuery = ref('')
 const activeCategory = ref('')
+const expandedPricingId = ref<number | null>(null)
+
+function togglePricingExpand(id: number) {
+	expandedPricingId.value = expandedPricingId.value === id ? null : id
+}
 
 const categories = [
 	{ value: '', label: '全部' },
@@ -89,12 +106,12 @@ const categoryLabel: Record<string, string> = {
 	rerank: '重排',
 }
 
-const categoryColor: Record<string, string> = {
-	chat: '#14b8a6',
-	embedding: '#8b5cf6',
-	image: '#f59e0b',
-	audio: '#10b981',
-	rerank: '#6b7280',
+const categoryBadgeStyle: Record<string, string> = {
+	chat: 'background:rgba(20,184,166,0.1);color:#0d9488',
+	embedding: 'background:rgba(139,92,246,0.1);color:#7c3aed',
+	image: 'background:rgba(245,158,11,0.1);color:#d97706',
+	audio: 'background:rgba(16,185,129,0.1);color:#059669',
+	rerank: 'background:rgba(107,114,128,0.1);color:#4b5563',
 }
 
 const filteredModels = computed(() => {
@@ -129,6 +146,36 @@ function formatTokens(n: number): string {
 	return String(n)
 }
 
+function formatPrice(n: number | null): string {
+	if (n == null || n === 0) return '-'
+	if (n >= 1) return '$' + n.toFixed(2)
+	if (n >= 0.01) return '$' + n.toFixed(3)
+	return '$' + n.toFixed(4)
+}
+
+function formatTokenRange(min: number, max: number | null): string {
+	const fmtMin = formatTokens(min)
+	if (max == null) return fmtMin + '+'
+	return fmtMin + ' ~ ' + formatTokens(max)
+}
+
+function hasPricing(m: ModelItem): boolean {
+	if (m.billing_mode === 'per_request' && m.per_request_price != null) return true
+	if (m.billing_mode === 'token' && (m.input_price != null || m.output_price != null)) return true
+	if (m.billing_mode === 'tiered' && m.pricing_tiers?.length > 0) return true
+	return false
+}
+
+function getTieredStartPrice(m: ModelItem): string {
+	if (!m.pricing_tiers?.length) return '-'
+	const first = m.pricing_tiers[0]
+	if (first.input_price > 0 && first.output_price > 0) {
+		return formatPrice(Math.min(first.input_price, first.output_price))
+	}
+	if (first.input_price > 0) return formatPrice(first.input_price)
+	return formatPrice(first.output_price)
+}
+
 async function fetchModels() {
 	loading.value = true
 	try {
@@ -225,74 +272,117 @@ onMounted(fetchModels)
 							</span>
 						</div>
 
-						<!-- Section 1: Category + Name -->
-						<div class="mb-3" :class="{ 'pr-14': m.discount_ratio && m.discount_ratio < 1 }">
-							<div class="flex items-center gap-2">
-								<span
-									class="shrink-0 px-2.5 py-0.5 text-[10px] font-medium text-white rounded-full"
-									:style="{ background: categoryColor[m.category] || '#6b7280' }"
-								>
-									{{ categoryLabel[m.category] || m.category }}
-								</span>
-								<h3 class="text-sm font-semibold text-gray-900 truncate">{{ m.model_name || m.model_id }}</h3>
-							</div>
-							<div class="flex items-center gap-1 mt-0.5">
-								<p class="text-xs text-gray-400 font-mono truncate">{{ m.model_id }}</p>
-								<button
-									class="shrink-0 p-0.5 rounded text-gray-300 hover:text-primary-500 transition-colors duration-150"
-									title="复制模型编码"
-									@click="copyModelId(m.model_id)"
-								>
-									<Icon v-if="copiedId !== m.model_id" name="copy" size="xs" />
-									<Icon v-else name="check" size="xs" class="text-emerald-500" />
-								</button>
-							</div>
+						<!-- Header: Category badge + Model name -->
+						<div class="flex items-center gap-2" :class="{ 'pr-14': m.discount_ratio && m.discount_ratio < 1 }">
+							<span
+								class="shrink-0 px-1.5 py-0.5 text-[10px] font-medium rounded"
+								:style="categoryBadgeStyle[m.category] || categoryBadgeStyle.rerank"
+							>
+								{{ categoryLabel[m.category] || m.category }}
+							</span>
+							<h3 class="text-sm font-semibold text-gray-900 truncate">{{ m.model_name || m.model_id }}</h3>
 						</div>
 
-						<!-- Section 2: Description (fixed height, always reserved) -->
-						<div class="mb-3 h-8">
-							<p v-if="m.description" class="text-xs text-gray-500 line-clamp-2">{{ m.description }}</p>
-							<span v-else class="text-xs text-gray-300">暂无描述</span>
+						<!-- Model ID + Copy -->
+						<div class="flex items-center gap-1 mt-1">
+							<span class="text-xs text-gray-400 font-mono truncate">{{ m.model_id }}</span>
+							<button
+								class="shrink-0 p-0.5 rounded text-gray-300 hover:text-primary-500 transition-colors duration-150"
+								title="复制模型编码"
+								@click="copyModelId(m.model_id)"
+							>
+								<Icon v-if="copiedId !== m.model_id" name="copy" size="xs" />
+								<Icon v-else name="check" size="xs" class="text-emerald-500" />
+							</button>
 						</div>
 
-						<!-- Section 3: Token info + Tags -->
-						<div class="flex items-center justify-between gap-2 text-xs text-gray-500 flex-1">
-							<div class="flex flex-wrap gap-3">
-								<span class="flex items-center gap-1">
-									<Icon name="document" size="xs" class="text-gray-300" />
-									{{ m.max_context_tokens ? formatTokens(m.max_context_tokens) : '-' }}
-								</span>
-								<span class="flex items-center gap-1">
-									<Icon name="arrowUp" size="xs" class="text-gray-300" />
-									{{ m.max_output_tokens ? formatTokens(m.max_output_tokens) : '-' }}
-								</span>
-								<span class="flex items-center gap-1">
-									<Icon name="chart" size="xs" class="text-gray-300" />
-									{{ m.max_concurrency ?? '-' }}
-								</span>
-							</div>
-							<div v-if="parseTags(m.tags).length" class="flex flex-wrap gap-1 justify-end">
-								<span
-									v-for="tag in parseTags(m.tags)"
-									:key="tag"
-									class="rounded bg-gray-100 px-1.5 py-0.5 text-[10px] font-medium text-gray-500"
-								>
-									{{ tag }}
-								</span>
-							</div>
+						<!-- Description (only if exists) -->
+						<p v-if="m.description" class="mt-2.5 text-xs text-gray-500 line-clamp-2 leading-relaxed">
+							{{ m.description }}
+						</p>
+
+						<!-- Key specs (dot-separated) -->
+						<div class="mt-2.5 flex items-center gap-1.5 text-[11px] text-gray-400">
+							<span v-if="m.max_context_tokens">{{ formatTokens(m.max_context_tokens) }} 上下文</span>
+							<span v-if="m.max_context_tokens && m.max_output_tokens" class="text-gray-200">·</span>
+							<span v-if="m.max_output_tokens">{{ formatTokens(m.max_output_tokens) }} 输出</span>
+							<template v-if="m.max_concurrency">
+								<span class="text-gray-200">·</span>
+								<span>并发 {{ m.max_concurrency }}</span>
+							</template>
 						</div>
 
-						<!-- Section 4: Capabilities (bottom, fixed height) -->
-						<div class="mt-3 pt-3 border-t border-gray-100 min-h-[28px]">
-							<div v-if="parseCapabilities(m.capabilities).length" class="flex flex-wrap gap-1">
-								<span
-									v-for="cap in parseCapabilities(m.capabilities)"
-									:key="cap"
-									class="inline-flex items-center rounded-md bg-primary-50 px-1.5 py-0.5 text-[10px] font-medium text-primary-600"
-								>
-									{{ capabilityLabel[cap] || cap }}
-								</span>
-							</div>
+						<!-- Capabilities (max 4 visible) -->
+						<div v-if="parseCapabilities(m.capabilities).length" class="mt-2.5 flex flex-wrap gap-1">
+							<span
+								v-for="cap in parseCapabilities(m.capabilities).slice(0, 4)"
+								:key="cap"
+								class="inline-flex items-center rounded-md bg-primary-50 px-1.5 py-0.5 text-[10px] font-medium text-primary-600"
+							>
+								{{ capabilityLabel[cap] || cap }}
+							</span>
+							<span
+								v-if="parseCapabilities(m.capabilities).length > 4"
+								class="inline-flex items-center rounded-md bg-gray-50 px-1.5 py-0.5 text-[10px] text-gray-400"
+							>
+								+{{ parseCapabilities(m.capabilities).length - 4 }}
+							</span>
+						</div>
+
+						<!-- Pricing (pushed to bottom) -->
+						<div v-if="hasPricing(m)" class="mt-auto pt-3 border-t border-gray-100">
+							<template v-if="m.billing_mode === 'token'">
+								<div class="flex items-baseline gap-3 text-xs">
+									<span>
+										<span class="text-gray-400">输入</span>
+										<span class="ml-1 font-semibold text-gray-800">{{ formatPrice(m.input_price) }}</span>
+									</span>
+									<span>
+										<span class="text-gray-400">输出</span>
+										<span class="ml-1 font-semibold text-gray-800">{{ formatPrice(m.output_price) }}</span>
+									</span>
+									<span class="text-gray-300">/1M tokens</span>
+								</div>
+							</template>
+
+							<template v-else-if="m.billing_mode === 'per_request'">
+								<div class="text-xs">
+									<span class="font-semibold text-gray-800">{{ formatPrice(m.per_request_price) }}</span>
+									<span class="text-gray-400"> /次</span>
+								</div>
+							</template>
+
+							<template v-else-if="m.billing_mode === 'tiered'">
+								<div class="flex items-center justify-between">
+									<div class="text-xs">
+										<span class="font-semibold text-gray-800">{{ getTieredStartPrice(m) }}</span>
+										<span class="text-gray-400"> 起 /1M tokens</span>
+									</div>
+									<button
+										class="flex items-center gap-0.5 text-xs text-primary-500 hover:text-primary-600 transition-colors"
+										@click.stop="togglePricingExpand(m.id)"
+									>
+										{{ expandedPricingId === m.id ? '收起' : '阶梯详情' }}
+										<Icon :name="expandedPricingId === m.id ? 'chevronUp' : 'chevronDown'" size="xs" />
+									</button>
+								</div>
+								<div v-if="expandedPricingId === m.id && m.pricing_tiers?.length" class="mt-2 bg-gray-50 rounded-lg p-2.5 space-y-1.5">
+									<div
+										v-for="(tier, idx) in m.pricing_tiers"
+										:key="idx"
+										class="flex items-center justify-between text-xs"
+									>
+										<span class="text-gray-500 font-mono">{{ formatTokenRange(tier.min_tokens, tier.max_tokens) }}</span>
+										<span>
+											<span class="text-gray-400">输入</span>
+											<span class="font-medium text-gray-700 ml-0.5">{{ formatPrice(tier.input_price) }}</span>
+											<span class="text-gray-300 mx-1">|</span>
+											<span class="text-gray-400">输出</span>
+											<span class="font-medium text-gray-700 ml-0.5">{{ formatPrice(tier.output_price) }}</span>
+										</span>
+									</div>
+								</div>
+							</template>
 						</div>
 					</div>
 				</div>