fix: improve OpenRouter model validation to support newer models

- Changed validation threshold from >1 to >10 models to avoid false negatives
- Added pattern matching for known model providers (OpenAI, Google, X-AI, etc.)
- Allows newer models like gpt-5.1, gemini-3-pro-preview, and grok-4.1-fast
- Added comprehensive tests for the new validation logic

Fixes #9597

webview-ui/src/utils/__tests__/validate.spec.ts

@@ -4,6 +4,20 @@ import { RouterModels } from "@roo/api"
 
 import { getModelValidationError, validateApiConfigurationExcludingModelErrors } from "../validate"
 
+// Mock i18next
+vi.mock("i18next", () => ({
+	default: {
+		t: vi.fn((key: string, options?: any) => {
+			// Return the key as-is for testing (this matches what the actual code expects)
+			if (options?.modelId) {
+				return key // For validation.modelAvailability, just return the key
+			}
+			return key
+		}),
+		language: "en",
+	},
+}))
+
 describe("Model Validation Functions", () => {
 	const mockRouterModels: RouterModels = {
 		openrouter: {
@@ -47,6 +61,25 @@ describe("Model Validation Functions", () => {
 		chutes: {},
 	}
 
+	// Mock router models with many models (>10) to trigger validation
+	const mockRouterModelsWithManyModels: RouterModels = {
+		...mockRouterModels,
+		openrouter: {
+			...mockRouterModels.openrouter,
+			"model-1": { maxTokens: 1000, contextWindow: 10000, supportsImages: false, supportsPromptCache: false },
+			"model-2": { maxTokens: 1000, contextWindow: 10000, supportsImages: false, supportsPromptCache: false },
+			"model-3": { maxTokens: 1000, contextWindow: 10000, supportsImages: false, supportsPromptCache: false },
+			"model-4": { maxTokens: 1000, contextWindow: 10000, supportsImages: false, supportsPromptCache: false },
+			"model-5": { maxTokens: 1000, contextWindow: 10000, supportsImages: false, supportsPromptCache: false },
+			"model-6": { maxTokens: 1000, contextWindow: 10000, supportsImages: false, supportsPromptCache: false },
+			"model-7": { maxTokens: 1000, contextWindow: 10000, supportsImages: false, supportsPromptCache: false },
+			"model-8": { maxTokens: 1000, contextWindow: 10000, supportsImages: false, supportsPromptCache: false },
+			"model-9": { maxTokens: 1000, contextWindow: 10000, supportsImages: false, supportsPromptCache: false },
+			"model-10": { maxTokens: 1000, contextWindow: 10000, supportsImages: false, supportsPromptCache: false },
+			"model-11": { maxTokens: 1000, contextWindow: 10000, supportsImages: false, supportsPromptCache: false },
+		},
+	}
+
 	const allowAllOrganization: OrganizationAllowList = {
 		allowAll: true,
 		providers: {},
@@ -73,14 +106,92 @@ describe("Model Validation Functions", () => {
 			expect(result).toBeUndefined()
 		})
 
-		it("returns error for invalid OpenRouter model", () => {
+		it("returns error for invalid OpenRouter model when model list has many models", () => {
 			const config: ProviderSettings = {
 				apiProvider: "openrouter",
 				openRouterModelId: "invalid-model",
 			}
 
+			// Should return error when we have many models (>10) in cache
+			const result = getModelValidationError(config, mockRouterModelsWithManyModels, allowAllOrganization)
+			expect(result).toBe("settings:validation.modelAvailability")
+		})
+
+		it("allows newer OpenRouter models like gpt-5.1 even when not in cache", () => {
+			const config: ProviderSettings = {
+				apiProvider: "openrouter",
+				openRouterModelId: "openai/gpt-5.1",
+			}
+
+			// Should allow gpt-5.1 even when not in the model list
+			const result = getModelValidationError(config, mockRouterModelsWithManyModels, allowAllOrganization)
+			expect(result).toBeUndefined()
+		})
+
+		it("allows newer OpenRouter models like gemini-3-pro-preview even when not in cache", () => {
+			const config: ProviderSettings = {
+				apiProvider: "openrouter",
+				openRouterModelId: "google/gemini-3-pro-preview",
+			}
+
+			// Should allow gemini-3 models even when not in the model list
+			const result = getModelValidationError(config, mockRouterModelsWithManyModels, allowAllOrganization)
+			expect(result).toBeUndefined()
+		})
+
+		it("allows newer OpenRouter models like grok-4.1-fast:free even when not in cache", () => {
+			const config: ProviderSettings = {
+				apiProvider: "openrouter",
+				openRouterModelId: "x-ai/grok-4.1-fast:free",
+			}
+
+			// Should allow grok-4 models even when not in the model list
+			const result = getModelValidationError(config, mockRouterModelsWithManyModels, allowAllOrganization)
+			expect(result).toBeUndefined()
+		})
+
+		it("allows known pattern models for OpenRouter even when not in cache", () => {
+			const testCases = [
+				"openai/gpt-5",
+				"openai/gpt-5.2",
+				"google/gemini-4",
+				"x-ai/grok-5",
+				"anthropic/claude-4-opus",
+				"meta-llama/llama-4",
+				"mistralai/mistral-large-2",
+			]
+
+			testCases.forEach((modelId) => {
+				const config: ProviderSettings = {
+					apiProvider: "openrouter",
+					openRouterModelId: modelId,
+				}
+
+				const result = getModelValidationError(config, mockRouterModelsWithManyModels, allowAllOrganization)
+				expect(result).toBeUndefined()
+			})
+		})
+
+		it("still validates unknown pattern models for OpenRouter", () => {
+			const config: ProviderSettings = {
+				apiProvider: "openrouter",
+				openRouterModelId: "unknown-provider/unknown-model",
+			}
+
+			// Should still validate models that don't match known patterns
+			const result = getModelValidationError(config, mockRouterModelsWithManyModels, allowAllOrganization)
+			expect(result).toBe("settings:validation.modelAvailability")
+		})
+
+		it("does not validate when model list is small (<=10 models)", () => {
+			const config: ProviderSettings = {
+				apiProvider: "openrouter",
+				openRouterModelId: "any-model-id",
+			}
+
+			// Should not validate when we have few models in cache (initial state)
 			const result = getModelValidationError(config, mockRouterModels, allowAllOrganization)
-			expect(result).toBe("validation.modelAvailability")
+			expect(result).toBeUndefined()
 		})
 
 		it("returns error for model not allowed by organization", () => {
@@ -130,7 +241,7 @@ describe("Model Validation Functions", () => {
 			}
 
 			const result = getModelValidationError(config, mockRouterModels, allowAllOrganization)
-			expect(result).toBe("validation.modelId")
+			expect(result).toBe("settings:validation.modelId")
 		})
 
 		it("handles undefined model IDs gracefully", () => {
@@ -140,7 +251,7 @@ describe("Model Validation Functions", () => {
 			}
 
 			const result = getModelValidationError(config, mockRouterModels, allowAllOrganization)
-			expect(result).toBe("validation.modelId")
+			expect(result).toBe("settings:validation.modelId")
 		})
 	})
 
@@ -164,7 +275,7 @@ describe("Model Validation Functions", () => {
 			}
 
 			const result = validateApiConfigurationExcludingModelErrors(config, mockRouterModels, allowAllOrganization)
-			expect(result).toBe("validation.apiKey")
+			expect(result).toBe("settings:validation.apiKey")
 		})
 
 		it("excludes model-specific errors", () => {
@@ -196,7 +307,7 @@ describe("Model Validation Functions", () => {
 		it("returns undefined for valid IO Intelligence model", () => {
 			const config: ProviderSettings = {
 				apiProvider: "io-intelligence",
-				glamaModelId: "valid-model",
+				ioIntelligenceModelId: "valid-model",
 			}
 
 			const result = getModelValidationError(config, mockRouterModels, allowAllOrganization)
@@ -206,7 +317,7 @@ describe("Model Validation Functions", () => {
 		it("returns error for invalid IO Intelligence model", () => {
 			const config: ProviderSettings = {
 				apiProvider: "io-intelligence",
-				glamaModelId: "invalid-model",
+				ioIntelligenceModelId: "invalid-model",
 			}
 
 			const result = getModelValidationError(config, mockRouterModels, allowAllOrganization)

webview-ui/src/utils/validate.ts

@@ -276,7 +276,31 @@ function validateDynamicProviderModelId(
 
 	const models = routerModels?.[provider]
 
-	if (models && Object.keys(models).length > 1 && !Object.keys(models).includes(modelId)) {
+	// Only validate if we have a substantial model list (more than 10 models)
+	// This prevents false negatives for newer models that haven't been added to the cache yet
+	// OpenRouter frequently adds new models that are immediately usable but may not appear
+	// in the /models endpoint right away
+	if (models && Object.keys(models).length > 10 && !Object.keys(models).includes(modelId)) {
+		// For OpenRouter specifically, we'll be more lenient with model validation
+		// to support newer models like gpt-5.1, gemini-3-pro-preview, grok-4.1-fast:free
+		if (provider === "openrouter") {
+			// Only show a warning for models that don't match any known pattern
+			// Allow models that follow common naming patterns even if not in cache
+			const knownPatterns = [
+				/^openai\/gpt-\d+(\.\d+)?/, // Matches gpt-5, gpt-5.1, etc.
+				/^google\/gemini-\d+/, // Matches gemini-3, gemini-3-pro, etc.
+				/^x-ai\/grok-\d+/, // Matches grok-4, grok-4.1, etc.
+				/^anthropic\/claude-/, // Claude models
+				/^meta-llama\//, // Meta Llama models
+				/^mistralai\//, // Mistral models
+			]
+
+			// If the model matches a known pattern, allow it
+			if (knownPatterns.some((pattern) => pattern.test(modelId))) {
+				return undefined
+			}
+		}
+
 		return i18next.t("settings:validation.modelAvailability", { modelId })
 	}