From 851bec843fc7db0ebfb5f673bbe6c87a941ba6ec Mon Sep 17 00:00:00 2001
From: MasterChiefCN <49179444+MasterChiefCN@users.noreply.github.com>
Date: Fri, 21 Aug 2026 11:57:36 +0800
Subject: [PATCH] =?UTF-8?q?fix(billing):=20=E5=89=8D=E5=8F=B0=E6=A8=A1?=
=?UTF-8?q?=E5=9E=8B=20-=20=E6=94=AF=E6=8C=81=E7=81=AB=E5=B1=B1=E6=96=B9?=
=?UTF-8?q?=E8=88=9F=E8=A7=86=E9=A2=91=20Token=20=E8=AE=A1=E8=B4=B9?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
---
CHANGELOG.md | 2 +
backend/internal/service/finance.go | 3 +
backend/internal/service/logical_models.go | 33 +++++++++--
.../internal/service/logical_models_test.go | 15 +++++
backend/internal/service/model_router.go | 6 ++
.../logical-models/logical-models-page.tsx | 57 ++++++++++++++-----
6 files changed, 98 insertions(+), 18 deletions(-)
diff --git a/CHANGELOG.md b/CHANGELOG.md
index c6497773d..d11294aab 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -2,6 +2,8 @@
## Unreleased
+- 前台火山方舟视频模型支持统一 Token 定价，并在混用其他视频协议或线路协议变更时阻止不兼容计费。
+
## v1.1.4
- 修复 `/canvas` 路由与静态模型目录同名导致的容器端口跳转，确保域名访问保持原有 HTTPS 地址。
diff --git a/backend/internal/service/finance.go b/backend/internal/service/finance.go
index 44dce5b9c..30e83c6fc 100644
--- a/backend/internal/service/finance.go
+++ b/backend/internal/service/finance.go
@@ -498,6 +498,9 @@ func (s *Service) newLogicalModelBillingOrder(userID string, task *model.Task, i
}
amount, err = creditAmount(logicalModel.UnitPriceMicrocredits, quantity, 10_000)
case "token":
+ if channelModel.Capability != capability || !supportsTokenBilling(capability, channelModel.Protocol) {
+ return nil, BadAuthRequest("当前供应线路不支持前台模型的 Token 计费方式")
+ }
pricing := &model.ChannelModel{InputTokenPriceMicrocredits: logicalModel.InputPriceMicrocredits, OutputTokenPriceMicrocredits: logicalModel.OutputPriceMicrocredits, CachedTokenPriceMicrocredits: logicalModel.CachedPriceMicrocredits}
amount, err = tokenEstimateAmount(pricing, tokenEstimate, 10_000)
quantity = tokenEstimate.InputTokens + tokenEstimate.OutputTokens
diff --git a/backend/internal/service/logical_models.go b/backend/internal/service/logical_models.go
index 8be0840da..6fd85ffd4 100644
--- a/backend/internal/service/logical_models.go
+++ b/backend/internal/service/logical_models.go
@@ -278,7 +278,8 @@ func (s *Service) buildAdminLogicalModel(item model.LogicalModel, graph *reposit
}
_, channelOK = systemChannelByID[channelModel.ChannelID]
structurallyAvailable := route.Enabled && route.Weight > 0 && channelModel.Enabled && channelOK
- available := structurallyAvailable && (item.PricePolicy != "channel" || channelModel.PriceConfigured)
+ billingAvailable := item.PricePolicy != "unified" || item.BillingMode != "token" || supportsTokenBilling(item.Capability, channelModel.Protocol)
+ available := structurallyAvailable && billingAvailable && (item.PricePolicy != "channel" || channelModel.PriceConfigured)
admin.Routes = append(admin.Routes, AdminLogicalRoute{ID: route.ID, ChannelModelID: channelModel.ID, ChannelID: channelModel.ChannelID, ChannelModelKey: channelModel.ModelKey, ChannelModelName: channelModel.DisplayName, Enabled: route.Enabled, Priority: route.Priority, Weight: route.Weight, Available: available, structurallyAvailable: structurallyAvailable, CapabilitySpec: capabilitySpec})
}
routeSpecs := make([]CapabilitySpec, 0, len(admin.Routes))
@@ -479,9 +480,6 @@ func (s *Service) logicalModelBundle(actor *model.User, id string, req LogicalMo
if pricePolicy == "unified" && billingMode == "per_second" && capability != "video" {
return nil, nil, nil, false, BadAuthRequest("只有视频前台模型可以按秒计费")
}
- if pricePolicy == "unified" && billingMode == "token" && capability != "text" {
- return nil, nil, nil, false, BadAuthRequest("当前仅文本前台模型支持 Token 计费")
- }
creating := strings.TrimSpace(id) == ""
var item *model.LogicalModel
if creating {
@@ -514,6 +512,7 @@ func (s *Service) logicalModelBundle(actor *model.User, id string, req LogicalMo
seenChannelModels := make(map[string]bool, len(req.Routes))
structuralRouteSpecs := make([]CapabilitySpec, 0, len(req.Routes))
settlementRouteSpecs := make([]CapabilitySpec, 0, len(req.Routes))
+ enabledRouteProtocols := make([]model.ChannelInterfaceType, 0, len(req.Routes))
for _, input := range req.Routes {
channelModelID := strings.TrimSpace(input.ChannelModelID)
if channelModelID == "" || seenChannelModels[channelModelID] {
@@ -531,6 +530,9 @@ func (s *Service) logicalModelBundle(actor *model.User, id string, req LogicalMo
if normalizeCapability(capabilitySpec.Capability) != capability {
return nil, nil, nil, false, BadAuthRequest("供应线路能力类型与前台模型不一致")
}
+ if input.Enabled {
+ enabledRouteProtocols = append(enabledRouteProtocols, channelModel.Protocol)
+ }
if req.Enabled && input.Enabled && input.Weight <= 0 {
return nil, nil, nil, false, BadAuthRequest("启用供应线路的同级权重必须大于 0")
}
@@ -552,6 +554,14 @@ func (s *Service) logicalModelBundle(actor *model.User, id string, req LogicalMo
}
routes = append(routes, model.LogicalModelRoute{ID: routeID, ChannelModelID: channelModel.ID, Enabled: input.Enabled, Priority: input.Priority, Weight: input.Weight, CreatedAt: time.Now(), UpdatedAt: time.Now()})
}
+ if pricePolicy == "unified" && billingMode == "token" {
+ if !supportsLogicalModelTokenBilling(capability, enabledRouteProtocols) {
+ return nil, nil, nil, false, BadAuthRequest("Token 计费仅支持文本前台模型，或全部启用供应线路均为火山方舟视频协议的视频前台模型")
+ }
+ if capability == "video" && req.OutputPriceMicrocredits <= 0 {
+ return nil, nil, nil, false, BadAuthRequest("火山方舟视频 Token 计费需要配置每百万视频 Token 价格")
+ }
+ }
// 停用必须始终可执行，便于管理员立即阻止失效线路继续对外服务；重新启用时再强校验结构能力和计费可用性。
if req.Enabled {
if len(structuralRouteSpecs) == 0 {
@@ -567,6 +577,21 @@ func (s *Service) logicalModelBundle(actor *model.User, id string, req LogicalMo
return item, revision, routes, creating, nil
}
+func supportsLogicalModelTokenBilling(capability string, enabledRouteProtocols []model.ChannelInterfaceType) bool {
+ if capability == "text" {
+ return true
+ }
+ if capability != "video" || len(enabledRouteProtocols) == 0 {
+ return false
+ }
+ for _, protocol := range enabledRouteProtocols {
+ if !supportsTokenBilling(capability, protocol) {
+ return false
+ }
+ }
+ return true
+}
+
func normalizeLogicalDefaults(spec CapabilitySpec, defaults map[string]any) (map[string]any, error) {
result := make(map[string]any, len(defaults))
for rawName, value := range defaults {
diff --git a/backend/internal/service/logical_models_test.go b/backend/internal/service/logical_models_test.go
index 748a22ee6..5af096129 100644
--- a/backend/internal/service/logical_models_test.go
+++ b/backend/internal/service/logical_models_test.go
@@ -150,6 +150,21 @@ func TestLogicalModelAvailabilityErrorRequiresSettlementReadyCoverage(t *testing
}
}
+func TestSupportsLogicalModelTokenBillingForArkVideoRoutes(t *testing.T) {
+ if !supportsLogicalModelTokenBilling("text", nil) {
+ t.Fatal("text logical models should support Token billing")
+ }
+ if !supportsLogicalModelTokenBilling("video", []model.ChannelInterfaceType{model.ChannelInterfaceVolcengineArkVideo}) {
+ t.Fatal("Ark video-only routes should support Token billing")
+ }
+ if supportsLogicalModelTokenBilling("video", nil) {
+ t.Fatal("video logical models without an enabled route must not support Token billing")
+ }
+ if supportsLogicalModelTokenBilling("video", []model.ChannelInterfaceType{model.ChannelInterfaceVolcengineArkVideo, model.ChannelInterfaceNewAPIVideo}) {
+ t.Fatal("mixed video protocols must not support Token billing")
+ }
+}
+
func TestChannelModelCapabilitySpecRequiresExplicitImageCapability(t *testing.T) {
channelModel := model.ChannelModel{Capability: "image", CapabilityConfigJSON: ""}
diff --git a/backend/internal/service/model_router.go b/backend/internal/service/model_router.go
index 624576157..e0e3db43f 100644
--- a/backend/internal/service/model_router.go
+++ b/backend/internal/service/model_router.go
@@ -537,6 +537,9 @@ func (s *Service) loadRouteCatalog() (*routeCatalogSnapshot, error) {
if !ok || !channelModel.Enabled || !enabledSystemChannels[channelModel.ChannelID] {
continue
}
+ if item.PricePolicy == "unified" && item.BillingMode == "token" && !supportsTokenBilling(item.Capability, channelModel.Protocol) {
+ continue
+ }
if item.PricePolicy == "channel" && !channelModel.PriceConfigured {
continue
}
@@ -840,6 +843,9 @@ func (s *Service) routedModelForTaskSelection(task *model.Task) (*RoutedModel, e
if logicalModel.PricePolicy == "channel" && !channelModel.PriceConfigured {
return nil, errors.New("任务使用的模型服务价格配置已失效")
}
+ if logicalModel.PricePolicy == "unified" && logicalModel.BillingMode == "token" && !supportsTokenBilling(logicalModel.Capability, channelModel.Protocol) {
+ return nil, errors.New("任务使用的模型服务不再支持当前 Token 计费配置")
+ }
routed := &RoutedModel{LogicalModel: *logicalModel, Revision: *revision, Route: *route, ChannelModel: *channelModel, Defaults: defaults}
return routed, nil
}
diff --git a/web/src/pages/admin/logical-models/logical-models-page.tsx b/web/src/pages/admin/logical-models/logical-models-page.tsx
index 9319f1aad..ee4260a5f 100644
--- a/web/src/pages/admin/logical-models/logical-models-page.tsx
+++ b/web/src/pages/admin/logical-models/logical-models-page.tsx
@@ -8,6 +8,7 @@ import { PaginationBar } from "@/components/layout/workspace-page";
import { ModelIconPicker, ModelLogo } from "@/components/model-logo";
import { CapabilityCardPicker } from "@/components/model-protocol-picker";
import { formatCredits } from "@/constant/credits";
+import { modelProtocolSupportsTokenBilling } from "@/lib/model-protocols";
import { AdminPageFrame } from "@/pages/admin/components/admin-shell";
import { AdminDataTable, AdminFilterChip, AdminRowActions, AdminStatusBadge, AdminTableEmpty } from "@/pages/admin/components/admin-ui";
import { listAdminChannels } from "@/services/api/auth";
@@ -416,7 +417,7 @@ export default function LogicalModelsPage() {
-
+
@@ -599,18 +600,38 @@ function logicalModelStatusTag(item: AdminLogicalModel) {
return ;
}
-function PricingFields() {
+function PricingFields({ channelModels }: { channelModels: ChannelModel[] }) {
const form = Form.useFormInstance();
const pricePolicy = Form.useWatch("pricePolicy") as LogicalModelMutation["pricePolicy"] | undefined;
const billingMode = Form.useWatch("billingMode") as LogicalModelMutation["billingMode"] | undefined;
const capability = Form.useWatch("capability") as CapabilityKind | undefined;
- const modes = [{ label: "按次", value: "fixed_request" }];
- if (capability === "video") modes.push({ label: "按秒", value: "per_second" });
+ const routes = (Form.useWatch("routes") || []) as RouteRuleRow[];
+ const enabledRoutes = routes.filter((route) => route.enabled);
+ const tokenBillingSupported =
+ capability === "text" ||
+ (capability === "video" &&
+ enabledRoutes.length > 0 &&
+ enabledRoutes.every((route) => {
+ const channelModel = channelModels.find((item) => item.id === route.channelModelId);
+ return modelProtocolSupportsTokenBilling(channelModel?.capability, channelModel?.protocol);
+ }));
+ const modes: Array<{ label: string; value: LogicalModelMutation["billingMode"]; disabled?: boolean }> = [{ label: "按次", value: "fixed_request" }];
+ if (capability === "video") {
+ modes.push({ label: "按秒", value: "per_second" });
+ modes.push({ label: "Token", value: "token", disabled: !tokenBillingSupported });
+ }
if (capability === "text") modes.push({ label: "Token", value: "token" });
+
+ useEffect(() => {
+ if (pricePolicy === "unified" && billingMode === "token" && !tokenBillingSupported) {
+ form.setFieldValue("billingMode", capability === "video" ? "per_second" : "fixed_request");
+ }
+ }, [billingMode, capability, form, pricePolicy, tokenBillingSupported]);
+
const changePolicy = (value: string | number) => {
const nextPolicy = value as LogicalModelMutation["pricePolicy"];
if (nextPolicy !== "unified") return;
- const supportedModes = modes.map((item) => item.value);
+ const supportedModes = modes.filter((item) => !item.disabled).map((item) => item.value);
if (!billingMode || !supportedModes.includes(billingMode)) {
form.setFieldValue("billingMode", capability === "text" ? "token" : capability === "video" ? "per_second" : "fixed_request");
}
@@ -639,18 +660,25 @@ function PricingFields() {
) : null}
+ {capability === "video" && !tokenBillingSupported ? Token 计费仅在所有启用供应线路都使用火山方舟视频协议时可用。
: null}
{billingMode === "token" ? (
-
-
+ capability === "video" ? (
+
-
-
-
-
-
-
-
+ ) : (
+
+
+
+
+
+
+
+
+
+
+
+ )
) : null}
) : (
@@ -666,6 +694,7 @@ function CreditsInput({ value = 0, onChange }: { value?: number; onChange?: (val
function logicalPriceLabel(item: AdminLogicalModel) {
if (item.pricePolicy === "channel") return 跟随供应价格;
+ if (item.billingMode === "token" && item.capability === "video") return 视频 {formatCredits(item.outputPriceMicrocredits)} / 百万 Token;
if (item.billingMode === "token")
return (