[Feat] Add GLM-5.3-Flash to Z AI

packages/types/src/__tests__/lite-llm.test.ts

@@ -21,6 +21,7 @@ describe("LiteLLM preserveReasoning model detection", () => {
 	it("matches case-insensitively", () => {
 		expect(isLiteLLMPreserveReasoningModel("MiniMax-M2.7-Highspeed")).toBe(true)
 		expect(isLiteLLMPreserveReasoningModel("GLM-5.3")).toBe(true)
+		expect(isLiteLLMPreserveReasoningModel("GLM-5.3-FLASH")).toBe(true)
 		expect(isLiteLLMPreserveReasoningModel("QWEN3.8-MAX")).toBe(true)
 	})
 

packages/types/src/providers/lite-llm.ts

@@ -55,6 +55,7 @@ export const LITELLM_PRESERVE_REASONING_MODEL_IDS = [
 	"glm-5.1",
 	"glm-5.2",
 	"glm-5.3",
+	"glm-5.3-flash",
 	"glm-5-turbo",
 
 	// bedrock.ts, minimax.ts, opencode-go.ts

packages/types/src/providers/zai.ts

@@ -6,6 +6,7 @@ import { ZaiApiLine } from "../provider-settings.js"
 // https://docs.z.ai/guides/llm/glm-4.5
 // https://docs.z.ai/guides/llm/glm-4.6
 // https://docs.z.ai/guides/llm/glm-5.3
+// https://docs.z.ai/guides/vlm/glm-5.3-flash
 // https://docs.z.ai/guides/llm/glm-5.1
 // https://docs.z.ai/guides/llm/glm-5-turbo
 // https://docs.z.ai/guides/overview/pricing
@@ -26,6 +27,13 @@ const glm53ModelInfo = {
 		"GLM-5.3 is Zhipu's flagship coding and agent model with a 1M context window, 128k max output, and always-on reasoning with configurable effort (Low/High/Max).",
 } as const satisfies ModelInfo
 
+const glm53FlashModelInfo = {
+	...glm53ModelInfo,
+	supportsImages: true,
+	description:
+		"GLM-5.3-Flash is Zhipu's efficient multimodal coding and agent model with a 1M context window, 128k max output, and always-on reasoning with configurable effort (Low/High/Max).",
+} as const satisfies ModelInfo
+
 export type InternationalZAiModelId = keyof typeof internationalZAiModels
 export const internationalZAiDefaultModelId: InternationalZAiModelId = "glm-4.7"
 export const internationalZAiModels = {
@@ -193,6 +201,13 @@ export const internationalZAiModels = {
 		cacheWritesPrice: 0,
 		cacheReadsPrice: 0.26,
 	},
+	"glm-5.3-flash": {
+		...glm53FlashModelInfo,
+		inputPrice: 0.15,
+		outputPrice: 0.5,
+		cacheWritesPrice: 0,
+		cacheReadsPrice: 0.03,
+	},
 	"glm-5-turbo": {
 		maxTokens: 131_072,
 		contextWindow: 202_752,
@@ -505,6 +520,13 @@ export const mainlandZAiCodingPlanOnlyModels = {
 		cacheWritesPrice: 0,
 		cacheReadsPrice: 0.13,
 	},
+	"glm-5.3-flash": {
+		...glm53FlashModelInfo,
+		inputPrice: 0.075,
+		outputPrice: 0.25,
+		cacheWritesPrice: 0,
+		cacheReadsPrice: 0.015,
+	},
 } as const satisfies Record<string, ModelInfo>
 
 export const ZAI_DEFAULT_TEMPERATURE = 0.6

src/api/providers/__tests__/zai.spec.ts

@@ -167,6 +167,31 @@ describe("ZAiHandler", () => {
 			expect(model.info.cacheReadsPrice).toBe(0.26)
 		})
 
+		it("should expose multimodal GLM-5.3-Flash for the international Coding Plan", () => {
+			const model = new ZAiHandler({
+				apiModelId: "glm-5.3-flash",
+				zaiApiKey: "test-zai-api-key",
+				zaiApiLine: "international_coding",
+			}).getModel()
+
+			expect(model.id).toBe("glm-5.3-flash")
+			expect(model.info).toMatchObject({
+				contextWindow: 1_000_000,
+				maxTokens: 131_072,
+				supportsImages: true,
+				supportsPromptCache: true,
+				supportsMaxTokens: true,
+				supportsReasoningEffort: ["low", "high", "max"],
+				requiredReasoningEffort: true,
+				reasoningEffort: "max",
+				preserveReasoning: true,
+				defaultTemperature: 1,
+				inputPrice: 0.15,
+				outputPrice: 0.5,
+				cacheReadsPrice: 0.03,
+			})
+		})
+
 		it("should return GLM-5-Turbo international model with thinking support", () => {
 			const testModelId: InternationalZAiModelId = "glm-5-turbo"
 			const handlerWithModel = new ZAiHandler({
@@ -319,6 +344,22 @@ describe("ZAiHandler", () => {
 			expect(model.info.cacheReadsPrice).toBe(0.13)
 		})
 
+		it("should expose GLM-5.3-Flash for the China Coding Plan", () => {
+			const model = new ZAiHandler({
+				apiModelId: "glm-5.3-flash",
+				zaiApiKey: "test-zai-api-key",
+				zaiApiLine: "china_coding",
+			}).getModel()
+
+			expect(model.id).toBe("glm-5.3-flash")
+			expect(model.info.supportsImages).toBe(true)
+			expect(model.info.supportsReasoningEffort).toEqual(["low", "high", "max"])
+			expect(model.info.requiredReasoningEffort).toBe(true)
+			expect(model.info.inputPrice).toBe(0.075)
+			expect(model.info.outputPrice).toBe(0.25)
+			expect(model.info.cacheReadsPrice).toBe(0.015)
+		})
+
 		it("should return GLM-4.7 China model with thinking support", () => {
 			const testModelId: MainlandZAiModelId = "glm-4.7"
 			const handlerWithModel = new ZAiHandler({
@@ -393,6 +434,7 @@ describe("ZAiHandler", () => {
 
 		it("should expose GLM-5.3 on the international API", () => {
 			expect(getZAiModels("international_api")).toHaveProperty("glm-5.3")
+			expect(getZAiModels("international_api")).toHaveProperty("glm-5.3-flash")
 			const handlerWithModel = new ZAiHandler({
 				apiModelId: "glm-5.3",
 				zaiApiKey: "test-zai-api-key",
@@ -442,6 +484,7 @@ describe("ZAiHandler", () => {
 
 		it("should not expose Coding Plan-only models", () => {
 			expect(getZAiModels("china_api")).not.toHaveProperty("glm-5.3")
+			expect(getZAiModels("china_api")).not.toHaveProperty("glm-5.3-flash")
 		})
 	})
 
@@ -692,6 +735,29 @@ describe("ZAiHandler", () => {
 			)
 		})
 
+		it("should use the official GLM-5.3-Flash thinking and sampling defaults", async () => {
+			const handlerWithModel = new ZAiHandler({
+				apiModelId: "glm-5.3-flash",
+				zaiApiKey: "test-zai-api-key",
+				zaiApiLine: "international_coding",
+				reasoningEffort: "disable",
+			})
+
+			mockCreate.mockImplementationOnce(() => asyncStreamFrom([]))
+
+			const messageGenerator = handlerWithModel.createMessage("system prompt", [])
+			await messageGenerator.next()
+
+			expect(mockCreate).toHaveBeenCalledWith(
+				expect.objectContaining({
+					model: "glm-5.3-flash",
+					thinking: { type: "enabled", clear_thinking: false },
+					reasoning_effort: "max",
+					temperature: 1,
+				}),
+			)
+		})
+
 		it("should keep GLM-5.3 reasoning enabled when the master reasoning setting is disabled", async () => {
 			const handlerWithModel = new ZAiHandler({
 				apiModelId: "glm-5.3",
@@ -734,6 +800,26 @@ describe("ZAiHandler", () => {
 			})
 		})
 
+		it("should use the official GLM-5.3-Flash parameters for completePrompt", async () => {
+			const handlerWithModel = new ZAiHandler({
+				apiModelId: "glm-5.3-flash",
+				zaiApiKey: "test-zai-api-key",
+				zaiApiLine: "international_api",
+				reasoningEffort: "low",
+			})
+
+			mockCreate.mockResolvedValueOnce({ choices: [{ message: { content: "response" } }] })
+
+			await expect(handlerWithModel.completePrompt("prompt")).resolves.toBe("response")
+			expect(mockCreate).toHaveBeenCalledWith({
+				model: "glm-5.3-flash",
+				messages: [{ role: "user", content: "prompt" }],
+				temperature: 1,
+				thinking: { type: "enabled", clear_thinking: false },
+				reasoning_effort: "low",
+			})
+		})
+
 		it("should omit reasoning_effort for GLM-5.2 when reasoningEffort is set to disable", async () => {
 			const handlerWithModel = new ZAiHandler({
 				apiModelId: "glm-5.2",

src/api/providers/fetchers/__tests__/litellm.spec.ts

@@ -765,7 +765,7 @@ describe("getLiteLLMModels", () => {
 				data: {
 					data: [
 						{
-							model_name: "glm-5.3",
+							model_name: "glm-5.3-flash",
 							model_info: {
 								max_tokens: 8192,
 								max_input_tokens: 128000,
@@ -782,7 +782,7 @@ describe("getLiteLLMModels", () => {
 
 			const result = await getLiteLLMModels("test-api-key", "http://localhost:4000")
 
-			expect(result["glm-5.3"]).toMatchObject({ preserveReasoning: true })
+			expect(result["glm-5.3-flash"]).toMatchObject({ preserveReasoning: true })
 		})
 
 		it("does not match a model id that merely contains a known family as a substring", async () => {

src/api/providers/zai.ts

@@ -27,6 +27,8 @@ type ZAiChatCompletionParams = Omit<OpenAI.Chat.ChatCompletionCreateParams, "rea
 	reasoning_effort?: "none" | "minimal" | "low" | "medium" | "high" | "xhigh" | "max"
 }
 
+const isGlm53Model = (model: string) => model === "glm-5.3" || model === "glm-5.3-flash"
+
 export class ZAiHandler extends BaseOpenAiCompatibleProvider<string> {
 	constructor(options: ApiHandlerOptions) {
 		const apiLine = options.zaiApiLine ?? "international_coding"
@@ -105,7 +107,7 @@ export class ZAiHandler extends BaseOpenAiCompatibleProvider<string> {
 			stream_options: { include_usage: true },
 			// Models with required reasoning stay enabled even when an old setting requests disable.
 			thinking: useReasoning
-				? { type: "enabled", ...(model === "glm-5.3" && { clear_thinking: false }) }
+				? { type: "enabled", ...(isGlm53Model(model) && { clear_thinking: false }) }
 				: { type: "disabled" },
 			reasoning_effort: reasoningEffort,
 			tools: this.convertToolsForOpenAI(metadata?.tools),
@@ -143,7 +145,7 @@ export class ZAiHandler extends BaseOpenAiCompatibleProvider<string> {
 
 	override async completePrompt(prompt: string, options?: CompletePromptOptions): Promise<string> {
 		const { id: model, info } = this.getModel()
-		if (model !== "glm-5.3") {
+		if (!isGlm53Model(model)) {
 			return super.completePrompt(prompt, options)
 		}
 

webview-ui/src/components/settings/utils/__tests__/providerModelConfig.spec.ts

@@ -206,7 +206,7 @@ describe("providerModelConfig", () => {
 			expect(Object.keys(models).length).toBe(0)
 		})
 
-		it("shows GLM-5.3 for international Z.ai API and Coding Plan entrypoints", () => {
+		it("shows GLM-5.3 models for international Z.ai API and Coding Plan entrypoints", () => {
 			const internationalCoding = getStaticModelsForProvider("zai", undefined, {
 				apiProvider: "zai",
 				zaiApiLine: "international_coding",
@@ -225,9 +225,13 @@ describe("providerModelConfig", () => {
 			})
 
 			expect(internationalCoding).toHaveProperty("glm-5.3")
+			expect(internationalCoding).toHaveProperty("glm-5.3-flash")
 			expect(chinaCoding).toHaveProperty("glm-5.3")
+			expect(chinaCoding).toHaveProperty("glm-5.3-flash")
 			expect(internationalApi).toHaveProperty("glm-5.3")
+			expect(internationalApi).toHaveProperty("glm-5.3-flash")
 			expect(chinaApi).not.toHaveProperty("glm-5.3")
+			expect(chinaApi).not.toHaveProperty("glm-5.3-flash")
 		})
 	})
 

webview-ui/src/components/ui/hooks/__tests__/useSelectedModel.spec.ts

@@ -1331,13 +1331,14 @@ describe("useSelectedModel", () => {
 		it("uses the International Coding catalog when no API line is configured", () => {
 			const apiConfiguration: ProviderSettings = {
 				apiProvider: providerIdentifiers.zai,
-				apiModelId: "glm-5.3",
+				apiModelId: "glm-5.3-flash",
 			}
 
 			const { result } = renderHook(() => useSelectedModel(apiConfiguration), { wrapper: createWrapper() })
 
-			expect(result.current.id).toBe("glm-5.3")
-			expect(result.current.info).toEqual(getZAiModels("international_coding")["glm-5.3"])
+			expect(result.current.id).toBe("glm-5.3-flash")
+			expect(result.current.info).toEqual(getZAiModels("international_coding")["glm-5.3-flash"])
+			expect(result.current.info?.supportsImages).toBe(true)
 		})
 
 		it("uses the China Coding catalog for GLM-5.3", () => {