lunanting/ai-chatbot
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix: require explicit multimodal flags
Some checks failed
Lint / build (20) (push) Has been cancelled
Details
Playwright Tests / test (push) Has been cancelled
Details

Browse Source
This commit is contained in:
Your Name 2026-03-19 00:46:47 +08:00
parent 7cd29fe2b5
commit d6a06e0cd0
4 changed files with 5 additions and 77 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
app/(chat)/api/chat/route.ts
View File

@ -175,6 +175,7 @@ import { convertToUIMessages, generateUUID } from "@/lib/utils";
import { isAiModelAgentAuto } from "@/lib/db/ai-model-flags";
import { generateTitleFromUserMessage } from "../../actions";
import { type PostRequestBody, postRequestBodySchema } from "./schema";
import { isAiModelMultimodal } from "@/lib/db/ai-model-flags";
/**
* 将图片 URL 转换为 base64 格式
@ -3206,7 +3207,7 @@ export async function POST(request: Request) {
: false));
const isConversationMultimodalModel =
!selectedModelConfig.isSubModel &&
selectedModelConfig.supportsVision &&
isAiModelMultimodal(selectedModelConfig) &&
!selectedModelConfig.supportsImageGen &&
!isVideoProviderType(selectedModelProviderType);
requestedModelIsFixedVisual = isFixedVisualModel;

33
lib/ai/auto-agent-graph.ts
View File

@ -964,38 +964,7 @@ function isAgentGenerateTextProviderType(providerType: string): boolean {
function isLikelyVisionCapableModel(item: EnabledModelWithProvider) {
if (!item.model.isEnabled || item.model.isSubModel) return false;
if (item.model.supportsVision) return true;
if (isAiModelMultimodal(item.model)) return true;
const providerType = String(item.provider.type || "");
const modelText = `${item.model.displayName || ""} ${item.model.modelId || ""}`.toLowerCase();
if (providerType === "volcano-multimodal") return true;
if (
providerType === "zhipu" &&
/(glm[-_.]?4v|glm[-_.]?4\.?6v|vision|多模态)/i.test(modelText)
) {
return true;
}
if (
providerType === "alibaba" &&
/(qwen|vl|vision|通义|多模态)/i.test(modelText)
) {
return true;
}
if (
["custom", "openai", "local"].includes(providerType) &&
/(doubao[-_.]?seed[-_.]?2[-_.]?0|seed[-_.]?2[-_.]?0|qwen[-_.]?vl|glm[-_.]?4v|gpt[-_.]?4o|vision|vl|多模态)/i.test(
modelText,
)
) {
return true;
}
return false;
return Boolean(item.model.supportsVision) || isAiModelMultimodal(item.model);
}
function getAgentModelIncompatibleReason(params: {

2
lib/ai/model-router.ts
View File

@ -78,7 +78,7 @@ function isConversationalCandidate(
const hasVision = hasVisionCapability(item.model);
if (matchesPreferredProvider) {
return requireVision ? hasVision || providerType === "volcano-multimodal" : true;
return requireVision ? hasVision : true;
}
if (IMAGE_ONLY_PROVIDER_TYPES.has(providerType)) {

44
lib/ai/smart-analyzer.ts
View File

@ -1686,50 +1686,8 @@ function isLikelyVisionCapableModel(modelWithProvider: {
};
}) {
const model = modelWithProvider.model;
const providerType = String(modelWithProvider.provider?.type || "");
if (!model?.isEnabled || model.isSubModel) return false;
if (isAiModelMultimodal(model)) return true;
const modelText = `${model.displayName || ""} ${model.modelId || ""}`.toLowerCase();
const likelyImageGenerationOnly =
!model.supportsVision &&
!isAiModelMultimodal(model) &&
Boolean(model.supportsImageGen) &&
((["custom", "openai", "local"].includes(providerType) &&
/(doubao[-_.]?seed[-_.]?2[-_.]?0|seed[-_.]?2[-_.]?0|seedream|gpt[-_.]?image|nano[-_.]?banana|flux|imagen|image[-_.]?edit)/i.test(
modelText,
)) ||
["gemini", "huanke-gemini", "alibaba-image", "ruxa-image", "fal-image"].includes(
providerType,
));
if (likelyImageGenerationOnly) return false;
if (model.supportsVision) return true;
if (providerType === "volcano-vision" || providerType === "volcano-multimodal") return true;
if (
providerType === "zhipu" &&
/(glm[-_.]?4v|glm[-_.]?4\.?6v|vision|多模态)/i.test(modelText)
) {
return true;
}
if (
providerType === "alibaba" &&
/(qwen|vl|vision|通义|多模态)/i.test(modelText)
) {
return true;
}
if (
["custom", "openai", "local"].includes(providerType) &&
/(doubao[-_.]?seed[-_.]?2[-_.]?0|seed[-_.]?2[-_.]?0|qwen[-_.]?vl|glm[-_.]?4v|gpt[-_.]?4o|vision|vl|多模态)/i.test(
modelText,
)
) {
return true;
}
return false;
return Boolean(model.supportsVision) || isAiModelMultimodal(model);
}
function getPreferredVolcanoMultimodalAnalyzerCandidate(