Merge pull request #658 from samhvw8/feat/roo-code-requesty-provider

feat: add Requesty API provider support

src/api/index.ts

@@ -15,6 +15,7 @@ import { MistralHandler } from "./providers/mistral"
 import { VsCodeLmHandler } from "./providers/vscode-lm"
 import { ApiStream } from "./transform/stream"
 import { UnboundHandler } from "./providers/unbound"
+import { RequestyHandler } from "./providers/requesty"
 
 export interface SingleCompletionHandler {
 	completePrompt(prompt: string): Promise<string>
@@ -56,6 +57,8 @@ export function buildApiHandler(configuration: ApiConfiguration): ApiHandler {
 			return new MistralHandler(options)
 		case "unbound":
 			return new UnboundHandler(options)
+		case "requesty":
+			return new RequestyHandler(options)
 		default:
 			return new AnthropicHandler(options)
 	}

src/api/providers/__tests__/requesty.test.ts

@@ -0,0 +1,247 @@
+import { Anthropic } from "@anthropic-ai/sdk"
+import OpenAI from "openai"
+import { ApiHandlerOptions, ModelInfo, requestyModelInfoSaneDefaults } from "../../../shared/api"
+import { RequestyHandler } from "../requesty"
+import { convertToOpenAiMessages } from "../../transform/openai-format"
+import { convertToR1Format } from "../../transform/r1-format"
+
+// Mock OpenAI and transform functions
+jest.mock("openai")
+jest.mock("../../transform/openai-format")
+jest.mock("../../transform/r1-format")
+
+describe("RequestyHandler", () => {
+	let handler: RequestyHandler
+	let mockCreate: jest.Mock
+
+	const defaultOptions: ApiHandlerOptions = {
+		requestyApiKey: "test-key",
+		requestyModelId: "test-model",
+		requestyModelInfo: {
+			maxTokens: 1000,
+			contextWindow: 4000,
+			supportsPromptCache: false,
+			supportsImages: true,
+			inputPrice: 0,
+			outputPrice: 0,
+		},
+		openAiStreamingEnabled: true,
+		includeMaxTokens: true, // Add this to match the implementation
+	}
+
+	beforeEach(() => {
+		// Clear mocks
+		jest.clearAllMocks()
+
+		// Setup mock create function
+		mockCreate = jest.fn()
+
+		// Mock OpenAI constructor
+		;(OpenAI as jest.MockedClass<typeof OpenAI>).mockImplementation(
+			() =>
+				({
+					chat: {
+						completions: {
+							create: mockCreate,
+						},
+					},
+				}) as unknown as OpenAI,
+		)
+
+		// Mock transform functions
+		;(convertToOpenAiMessages as jest.Mock).mockImplementation((messages) => messages)
+		;(convertToR1Format as jest.Mock).mockImplementation((messages) => messages)
+
+		// Create handler instance
+		handler = new RequestyHandler(defaultOptions)
+	})
+
+	describe("constructor", () => {
+		it("should initialize with correct options", () => {
+			expect(OpenAI).toHaveBeenCalledWith({
+				baseURL: "https://router.requesty.ai/v1",
+				apiKey: defaultOptions.requestyApiKey,
+				defaultHeaders: {
+					"HTTP-Referer": "https://github.com/RooVetGit/Roo-Cline",
+					"X-Title": "Roo Code",
+				},
+			})
+		})
+	})
+
+	describe("createMessage", () => {
+		const systemPrompt = "You are a helpful assistant"
+		const messages: Anthropic.Messages.MessageParam[] = [{ role: "user", content: "Hello" }]
+
+		describe("with streaming enabled", () => {
+			beforeEach(() => {
+				const stream = {
+					[Symbol.asyncIterator]: async function* () {
+						yield {
+							choices: [{ delta: { content: "Hello" } }],
+						}
+						yield {
+							choices: [{ delta: { content: " world" } }],
+							usage: {
+								prompt_tokens: 10,
+								completion_tokens: 5,
+							},
+						}
+					},
+				}
+				mockCreate.mockResolvedValue(stream)
+			})
+
+			it("should handle streaming response correctly", async () => {
+				const stream = handler.createMessage(systemPrompt, messages)
+				const results = []
+
+				for await (const chunk of stream) {
+					results.push(chunk)
+				}
+
+				expect(results).toEqual([
+					{ type: "text", text: "Hello" },
+					{ type: "text", text: " world" },
+					{
+						type: "usage",
+						inputTokens: 10,
+						outputTokens: 5,
+						cacheWriteTokens: undefined,
+						cacheReadTokens: undefined,
+					},
+				])
+
+				expect(mockCreate).toHaveBeenCalledWith({
+					model: defaultOptions.requestyModelId,
+					temperature: 0,
+					messages: [
+						{ role: "system", content: systemPrompt },
+						{ role: "user", content: "Hello" },
+					],
+					stream: true,
+					stream_options: { include_usage: true },
+					max_tokens: defaultOptions.requestyModelInfo?.maxTokens,
+				})
+			})
+
+			it("should not include max_tokens when includeMaxTokens is false", async () => {
+				handler = new RequestyHandler({
+					...defaultOptions,
+					includeMaxTokens: false,
+				})
+
+				await handler.createMessage(systemPrompt, messages).next()
+
+				expect(mockCreate).toHaveBeenCalledWith(
+					expect.not.objectContaining({
+						max_tokens: expect.any(Number),
+					}),
+				)
+			})
+
+			it("should handle deepseek-reasoner model format", async () => {
+				handler = new RequestyHandler({
+					...defaultOptions,
+					requestyModelId: "deepseek-reasoner",
+				})
+
+				await handler.createMessage(systemPrompt, messages).next()
+
+				expect(convertToR1Format).toHaveBeenCalledWith([{ role: "user", content: systemPrompt }, ...messages])
+			})
+		})
+
+		describe("with streaming disabled", () => {
+			beforeEach(() => {
+				handler = new RequestyHandler({
+					...defaultOptions,
+					openAiStreamingEnabled: false,
+				})
+
+				mockCreate.mockResolvedValue({
+					choices: [{ message: { content: "Hello world" } }],
+					usage: {
+						prompt_tokens: 10,
+						completion_tokens: 5,
+					},
+				})
+			})
+
+			it("should handle non-streaming response correctly", async () => {
+				const stream = handler.createMessage(systemPrompt, messages)
+				const results = []
+
+				for await (const chunk of stream) {
+					results.push(chunk)
+				}
+
+				expect(results).toEqual([
+					{ type: "text", text: "Hello world" },
+					{
+						type: "usage",
+						inputTokens: 10,
+						outputTokens: 5,
+					},
+				])
+
+				expect(mockCreate).toHaveBeenCalledWith({
+					model: defaultOptions.requestyModelId,
+					messages: [
+						{ role: "user", content: systemPrompt },
+						{ role: "user", content: "Hello" },
+					],
+				})
+			})
+		})
+	})
+
+	describe("getModel", () => {
+		it("should return correct model information", () => {
+			const result = handler.getModel()
+			expect(result).toEqual({
+				id: defaultOptions.requestyModelId,
+				info: defaultOptions.requestyModelInfo,
+			})
+		})
+
+		it("should use sane defaults when no model info provided", () => {
+			handler = new RequestyHandler({
+				...defaultOptions,
+				requestyModelInfo: undefined,
+			})
+
+			const result = handler.getModel()
+			expect(result).toEqual({
+				id: defaultOptions.requestyModelId,
+				info: requestyModelInfoSaneDefaults,
+			})
+		})
+	})
+
+	describe("completePrompt", () => {
+		beforeEach(() => {
+			mockCreate.mockResolvedValue({
+				choices: [{ message: { content: "Completed response" } }],
+			})
+		})
+
+		it("should complete prompt successfully", async () => {
+			const result = await handler.completePrompt("Test prompt")
+			expect(result).toBe("Completed response")
+			expect(mockCreate).toHaveBeenCalledWith({
+				model: defaultOptions.requestyModelId,
+				messages: [{ role: "user", content: "Test prompt" }],
+			})
+		})
+
+		it("should handle errors correctly", async () => {
+			const errorMessage = "API error"
+			mockCreate.mockRejectedValue(new Error(errorMessage))
+
+			await expect(handler.completePrompt("Test prompt")).rejects.toThrow(
+				`OpenAI completion error: ${errorMessage}`,
+			)
+		})
+	})
+})

src/api/providers/deepseek.ts

@@ -1,9 +1,9 @@
-import { OpenAiHandler } from "./openai"
-import { ApiHandlerOptions, ModelInfo } from "../../shared/api"
+import { OpenAiHandler, OpenAiHandlerOptions } from "./openai"
+import { ModelInfo } from "../../shared/api"
 import { deepSeekModels, deepSeekDefaultModelId } from "../../shared/api"
 
 export class DeepSeekHandler extends OpenAiHandler {
-	constructor(options: ApiHandlerOptions) {
+	constructor(options: OpenAiHandlerOptions) {
 		super({
 			...options,
 			openAiApiKey: options.deepSeekApiKey ?? "not-provided",

src/api/providers/openai.ts

@@ -11,16 +11,20 @@ import { ApiHandler, SingleCompletionHandler } from "../index"
 import { convertToOpenAiMessages } from "../transform/openai-format"
 import { convertToR1Format } from "../transform/r1-format"
 import { convertToSimpleMessages } from "../transform/simple-format"
-import { ApiStream } from "../transform/stream"
+import { ApiStream, ApiStreamUsageChunk } from "../transform/stream"
+
+export interface OpenAiHandlerOptions extends ApiHandlerOptions {
+	defaultHeaders?: Record<string, string>
+}
 
 export const DEEP_SEEK_DEFAULT_TEMPERATURE = 0.6
 const OPENAI_DEFAULT_TEMPERATURE = 0
 
 export class OpenAiHandler implements ApiHandler, SingleCompletionHandler {
-	protected options: ApiHandlerOptions
+	protected options: OpenAiHandlerOptions
 	private client: OpenAI
 
-	constructor(options: ApiHandlerOptions) {
+	constructor(options: OpenAiHandlerOptions) {
 		this.options = options
 
 		const baseURL = this.options.openAiBaseUrl ?? "https://api.openai.com/v1"
@@ -44,7 +48,7 @@ export class OpenAiHandler implements ApiHandler, SingleCompletionHandler {
 				apiVersion: this.options.azureApiVersion || azureOpenAiDefaultApiVersion,
 			})
 		} else {
-			this.client = new OpenAI({ baseURL, apiKey })
+			this.client = new OpenAI({ baseURL, apiKey, defaultHeaders: this.options.defaultHeaders })
 		}
 	}
 
@@ -103,11 +107,7 @@ export class OpenAiHandler implements ApiHandler, SingleCompletionHandler {
 					}
 				}
 				if (chunk.usage) {
-					yield {
-						type: "usage",
-						inputTokens: chunk.usage.prompt_tokens || 0,
-						outputTokens: chunk.usage.completion_tokens || 0,
-					}
+					yield this.processUsageMetrics(chunk.usage)
 				}
 			}
 		} else {
@@ -130,11 +130,15 @@ export class OpenAiHandler implements ApiHandler, SingleCompletionHandler {
 				type: "text",
 				text: response.choices[0]?.message.content || "",
 			}
-			yield {
-				type: "usage",
-				inputTokens: response.usage?.prompt_tokens || 0,
-				outputTokens: response.usage?.completion_tokens || 0,
-			}
+			yield this.processUsageMetrics(response.usage)
+		}
+	}
+
+	protected processUsageMetrics(usage: any): ApiStreamUsageChunk {
+		return {
+			type: "usage",
+			inputTokens: usage?.prompt_tokens || 0,
+			outputTokens: usage?.completion_tokens || 0,
 		}
 	}
 

src/api/providers/requesty.ts

@@ -0,0 +1,40 @@
+import { OpenAiHandler, OpenAiHandlerOptions } from "./openai"
+import { ModelInfo, requestyModelInfoSaneDefaults, requestyDefaultModelId } from "../../shared/api"
+import { ApiStream, ApiStreamUsageChunk } from "../transform/stream"
+
+export class RequestyHandler extends OpenAiHandler {
+	constructor(options: OpenAiHandlerOptions) {
+		if (!options.requestyApiKey) {
+			throw new Error("Requesty API key is required. Please provide it in the settings.")
+		}
+		super({
+			...options,
+			openAiApiKey: options.requestyApiKey,
+			openAiModelId: options.requestyModelId ?? requestyDefaultModelId,
+			openAiBaseUrl: "https://router.requesty.ai/v1",
+			openAiCustomModelInfo: options.requestyModelInfo ?? requestyModelInfoSaneDefaults,
+			defaultHeaders: {
+				"HTTP-Referer": "https://github.com/RooVetGit/Roo-Cline",
+				"X-Title": "Roo Code",
+			},
+		})
+	}
+
+	override getModel(): { id: string; info: ModelInfo } {
+		const modelId = this.options.requestyModelId ?? requestyDefaultModelId
+		return {
+			id: modelId,
+			info: this.options.requestyModelInfo ?? requestyModelInfoSaneDefaults,
+		}
+	}
+
+	protected override processUsageMetrics(usage: any): ApiStreamUsageChunk {
+		return {
+			type: "usage",
+			inputTokens: usage?.prompt_tokens || 0,
+			outputTokens: usage?.completion_tokens || 0,
+			cacheWriteTokens: usage?.cache_creation_input_tokens,
+			cacheReadTokens: usage?.cache_read_input_tokens,
+		}
+	}
+}

src/core/webview/ClineProvider.ts

@@ -59,6 +59,7 @@ type SecretKey =
 	| "deepSeekApiKey"
 	| "mistralApiKey"
 	| "unboundApiKey"
+	| "requestyApiKey"
 type GlobalStateKey =
 	| "apiProvider"
 	| "apiModelId"
@@ -122,6 +123,8 @@ type GlobalStateKey =
 	| "autoApprovalEnabled"
 	| "customModes" // Array of custom modes
 	| "unboundModelId"
+	| "requestyModelId"
+	| "requestyModelInfo"
 	| "unboundModelInfo"
 	| "modelTemperature"
 
@@ -130,6 +133,7 @@ export const GlobalFileNames = {
 	uiMessages: "ui_messages.json",
 	glamaModels: "glama_models.json",
 	openRouterModels: "openrouter_models.json",
+	requestyModels: "requesty_models.json",
 	mcpSettings: "cline_mcp_settings.json",
 	unboundModels: "unbound_models.json",
 }
@@ -686,6 +690,25 @@ export class ClineProvider implements vscode.WebviewViewProvider {
 							}
 						})
 
+						this.readRequestyModels().then((requestyModels) => {
+							if (requestyModels) {
+								this.postMessageToWebview({ type: "requestyModels", requestyModels })
+							}
+						})
+						this.refreshRequestyModels().then(async (requestyModels) => {
+							if (requestyModels) {
+								// update model info in state (this needs to be done here since we don't want to update state while settings is open, and we may refresh models there)
+								const { apiConfiguration } = await this.getState()
+								if (apiConfiguration.requestyModelId) {
+									await this.updateGlobalState(
+										"requestyModelInfo",
+										requestyModels[apiConfiguration.requestyModelId],
+									)
+									await this.postStateToWebview()
+								}
+							}
+						})
+
 						this.configManager
 							.listConfig()
 							.then(async (listApiConfig) => {
@@ -848,6 +871,12 @@ export class ClineProvider implements vscode.WebviewViewProvider {
 					case "refreshUnboundModels":
 						await this.refreshUnboundModels()
 						break
+					case "refreshRequestyModels":
+						if (message?.values?.apiKey) {
+							const requestyModels = await this.refreshRequestyModels(message?.values?.apiKey)
+							this.postMessageToWebview({ type: "requestyModels", requestyModels: requestyModels })
+						}
+						break
 					case "openImage":
 						openImage(message.text!)
 						break
@@ -1588,6 +1617,9 @@ export class ClineProvider implements vscode.WebviewViewProvider {
 			unboundApiKey,
 			unboundModelId,
 			unboundModelInfo,
+			requestyApiKey,
+			requestyModelId,
+			requestyModelInfo,
 			modelTemperature,
 		} = apiConfiguration
 		await this.updateGlobalState("apiProvider", apiProvider)
@@ -1630,6 +1662,9 @@ export class ClineProvider implements vscode.WebviewViewProvider {
 		await this.storeSecret("unboundApiKey", unboundApiKey)
 		await this.updateGlobalState("unboundModelId", unboundModelId)
 		await this.updateGlobalState("unboundModelInfo", unboundModelInfo)
+		await this.storeSecret("requestyApiKey", requestyApiKey)
+		await this.updateGlobalState("requestyModelId", requestyModelId)
+		await this.updateGlobalState("requestyModelInfo", requestyModelInfo)
 		await this.updateGlobalState("modelTemperature", modelTemperature)
 		if (this.cline) {
 			this.cline.api = buildApiHandler(apiConfiguration)
@@ -1773,6 +1808,93 @@ export class ClineProvider implements vscode.WebviewViewProvider {
 		}
 	}
 
+	// Requesty
+	async readRequestyModels(): Promise<Record<string, ModelInfo> | undefined> {
+		const requestyModelsFilePath = path.join(
+			await this.ensureCacheDirectoryExists(),
+			GlobalFileNames.requestyModels,
+		)
+		const fileExists = await fileExistsAtPath(requestyModelsFilePath)
+		if (fileExists) {
+			const fileContents = await fs.readFile(requestyModelsFilePath, "utf8")
+			return JSON.parse(fileContents)
+		}
+		return undefined
+	}
+
+	async refreshRequestyModels(apiKey?: string) {
+		const requestyModelsFilePath = path.join(
+			await this.ensureCacheDirectoryExists(),
+			GlobalFileNames.requestyModels,
+		)
+
+		const models: Record<string, ModelInfo> = {}
+		try {
+			const config: Record<string, any> = {}
+			if (!apiKey) {
+				apiKey = (await this.getSecret("requestyApiKey")) as string
+			}
+			if (apiKey) {
+				config["headers"] = { Authorization: `Bearer ${apiKey}` }
+			}
+
+			const response = await axios.get("https://router.requesty.ai/v1/models", config)
+			/*
+				{
+					"id": "anthropic/claude-3-5-sonnet-20240620",
+					"object": "model",
+					"created": 1738243330,
+					"owned_by": "system",
+					"input_price": 0.000003,
+					"caching_price": 0.00000375,
+					"cached_price": 3E-7,
+					"output_price": 0.000015,
+					"max_output_tokens": 8192,
+					"context_window": 200000,
+					"supports_caching": true,
+					"description": "Anthropic's most intelligent model. Highest level of intelligence and capability"
+					},
+				}
+			*/
+			if (response.data) {
+				const rawModels = response.data.data
+				const parsePrice = (price: any) => {
+					if (price) {
+						return parseFloat(price) * 1_000_000
+					}
+					return undefined
+				}
+				for (const rawModel of rawModels) {
+					const modelInfo: ModelInfo = {
+						maxTokens: rawModel.max_output_tokens,
+						contextWindow: rawModel.context_window,
+						supportsImages: rawModel.support_image,
+						supportsComputerUse: rawModel.support_computer_use,
+						supportsPromptCache: rawModel.supports_caching,
+						inputPrice: parsePrice(rawModel.input_price),
+						outputPrice: parsePrice(rawModel.output_price),
+						description: rawModel.description,
+						cacheWritesPrice: parsePrice(rawModel.caching_price),
+						cacheReadsPrice: parsePrice(rawModel.cached_price),
+					}
+
+					models[rawModel.id] = modelInfo
+				}
+			} else {
+				this.outputChannel.appendLine("Invalid response from Requesty API")
+			}
+			await fs.writeFile(requestyModelsFilePath, JSON.stringify(models))
+			this.outputChannel.appendLine(`Requesty models fetched and saved: ${JSON.stringify(models, null, 2)}`)
+		} catch (error) {
+			this.outputChannel.appendLine(
+				`Error fetching Requesty models: ${JSON.stringify(error, Object.getOwnPropertyNames(error), 2)}`,
+			)
+		}
+
+		await this.postMessageToWebview({ type: "requestyModels", requestyModels: models })
+		return models
+	}
+
 	// OpenRouter
 
 	async handleOpenRouterCallback(code: string) {
@@ -2391,6 +2513,9 @@ export class ClineProvider implements vscode.WebviewViewProvider {
 			unboundApiKey,
 			unboundModelId,
 			unboundModelInfo,
+			requestyApiKey,
+			requestyModelId,
+			requestyModelInfo,
 			modelTemperature,
 		] = await Promise.all([
 			this.getGlobalState("apiProvider") as Promise<ApiProvider | undefined>,
@@ -2468,6 +2593,9 @@ export class ClineProvider implements vscode.WebviewViewProvider {
 			this.getSecret("unboundApiKey") as Promise<string | undefined>,
 			this.getGlobalState("unboundModelId") as Promise<string | undefined>,
 			this.getGlobalState("unboundModelInfo") as Promise<ModelInfo | undefined>,
+			this.getSecret("requestyApiKey") as Promise<string | undefined>,
+			this.getGlobalState("requestyModelId") as Promise<string | undefined>,
+			this.getGlobalState("requestyModelInfo") as Promise<ModelInfo | undefined>,
 			this.getGlobalState("modelTemperature") as Promise<number | undefined>,
 		])
 
@@ -2527,6 +2655,9 @@ export class ClineProvider implements vscode.WebviewViewProvider {
 				unboundApiKey,
 				unboundModelId,
 				unboundModelInfo,
+				requestyApiKey,
+				requestyModelId,
+				requestyModelInfo,
 				modelTemperature,
 			},
 			lastShownAnnouncementId,
@@ -2681,6 +2812,7 @@ export class ClineProvider implements vscode.WebviewViewProvider {
 			"deepSeekApiKey",
 			"mistralApiKey",
 			"unboundApiKey",
+			"requestyApiKey",
 		]
 		for (const key of secretKeys) {
 			await this.storeSecret(key, undefined)

src/shared/ExtensionMessage.ts

@@ -30,6 +30,7 @@ export interface ExtensionMessage {
 		| "glamaModels"
 		| "openRouterModels"
 		| "openAiModels"
+		| "requestyModels"
 		| "mcpServers"
 		| "enhancedPrompt"
 		| "commitSearchResults"
@@ -67,6 +68,7 @@ export interface ExtensionMessage {
 	}>
 	partialMessage?: ClineMessage
 	glamaModels?: Record<string, ModelInfo>
+	requestyModels?: Record<string, ModelInfo>
 	openRouterModels?: Record<string, ModelInfo>
 	openAiModels?: string[]
 	unboundModels?: Record<string, ModelInfo>

src/shared/WebviewMessage.ts

@@ -43,6 +43,7 @@ export interface WebviewMessage {
 		| "refreshOpenRouterModels"
 		| "refreshOpenAiModels"
 		| "refreshUnboundModels"
+		| "refreshRequestyModels"
 		| "alwaysAllowBrowser"
 		| "alwaysAllowMcp"
 		| "alwaysAllowModeSwitch"

src/shared/__tests__/checkExistApiConfig.test.ts

@@ -51,6 +51,8 @@ describe("checkExistKey", () => {
 			deepSeekApiKey: undefined,
 			mistralApiKey: undefined,
 			vsCodeLmModelSelector: undefined,
+			requestyApiKey: undefined,
+			unboundApiKey: undefined,
 		}
 		expect(checkExistKey(config)).toBe(false)
 	})

src/shared/api.ts

@@ -15,6 +15,7 @@ export type ApiProvider =
 	| "vscode-lm"
 	| "mistral"
 	| "unbound"
+	| "requesty"
 
 export interface ApiHandlerOptions {
 	apiModelId?: string
@@ -61,6 +62,9 @@ export interface ApiHandlerOptions {
 	unboundApiKey?: string
 	unboundModelId?: string
 	unboundModelInfo?: ModelInfo
+	requestyApiKey?: string
+	requestyModelId?: string
+	requestyModelInfo?: ModelInfo
 	modelTemperature?: number
 }
 
@@ -354,6 +358,21 @@ export const glamaDefaultModelInfo: ModelInfo = {
 		"The new Claude 3.5 Sonnet delivers better-than-Opus capabilities, faster-than-Sonnet speeds, at the same Sonnet prices. Sonnet is particularly good at:\n\n- Coding: New Sonnet scores ~49% on SWE-Bench Verified, higher than the last best score, and without any fancy prompt scaffolding\n- Data science: Augments human data science expertise; navigates unstructured data while using multiple tools for insights\n- Visual processing: excelling at interpreting charts, graphs, and images, accurately transcribing text to derive insights beyond just the text alone\n- Agentic tasks: exceptional tool use, making it great at agentic tasks (i.e. complex, multi-step problem solving tasks that require engaging with other systems)\n\n#multimodal\n\n_This is a faster endpoint, made available in collaboration with Anthropic, that is self-moderated: response moderation happens on the provider's side instead of OpenRouter's. For requests that pass moderation, it's identical to the [Standard](/anthropic/claude-3.5-sonnet) variant._",
 }
 
+export const requestyDefaultModelInfo: ModelInfo = {
+	maxTokens: 8192,
+	contextWindow: 200_000,
+	supportsImages: true,
+	supportsComputerUse: true,
+	supportsPromptCache: true,
+	inputPrice: 3.0,
+	outputPrice: 15.0,
+	cacheWritesPrice: 3.75,
+	cacheReadsPrice: 0.3,
+	description:
+		"The new Claude 3.5 Sonnet delivers better-than-Opus capabilities, faster-than-Sonnet speeds, at the same Sonnet prices. Sonnet is particularly good at:\n\n- Coding: New Sonnet scores ~49% on SWE-Bench Verified, higher than the last best score, and without any fancy prompt scaffolding\n- Data science: Augments human data science expertise; navigates unstructured data while using multiple tools for insights\n- Visual processing: excelling at interpreting charts, graphs, and images, accurately transcribing text to derive insights beyond just the text alone\n- Agentic tasks: exceptional tool use, making it great at agentic tasks (i.e. complex, multi-step problem solving tasks that require engaging with other systems)\n\n#multimodal\n\n_This is a faster endpoint, made available in collaboration with Anthropic, that is self-moderated: response moderation happens on the provider's side instead of OpenRouter's. For requests that pass moderation, it's identical to the [Standard](/anthropic/claude-3.5-sonnet) variant._",
+}
+export const requestyDefaultModelId = "anthropic/claude-3-5-sonnet"
+
 // OpenRouter
 // https://openrouter.ai/models?order=newest&supported_parameters=tools
 export const openRouterDefaultModelId = "anthropic/claude-3.5-sonnet:beta" // will always exist in openRouterModels
@@ -428,6 +447,15 @@ export const openAiModelInfoSaneDefaults: ModelInfo = {
 	outputPrice: 0,
 }
 
+export const requestyModelInfoSaneDefaults: ModelInfo = {
+	maxTokens: -1,
+	contextWindow: 128_000,
+	supportsImages: true,
+	supportsPromptCache: false,
+	inputPrice: 0,
+	outputPrice: 0,
+}
+
 // Gemini
 // https://ai.google.dev/gemini-api/docs/models/gemini
 export type GeminiModelId = keyof typeof geminiModels

src/shared/checkExistApiConfig.ts

@@ -16,6 +16,8 @@ export function checkExistKey(config: ApiConfiguration | undefined) {
 				config.deepSeekApiKey,
 				config.mistralApiKey,
 				config.vsCodeLmModelSelector,
+				config.requestyApiKey,
+				config.unboundApiKey,
 			].some((key) => key !== undefined)
 		: false
 }

webview-ui/src/components/settings/ApiOptions.tsx

@@ -30,6 +30,8 @@ import {
 	vertexModels,
 	unboundDefaultModelId,
 	unboundDefaultModelInfo,
+	requestyDefaultModelId,
+	requestyDefaultModelInfo,
 } from "../../../../src/shared/api"
 import { ExtensionMessage } from "../../../../src/shared/ExtensionMessage"
 import { useExtensionState } from "../../context/ExtensionStateContext"
@@ -41,6 +43,7 @@ import { GlamaModelPicker } from "./GlamaModelPicker"
 import { UnboundModelPicker } from "./UnboundModelPicker"
 import { ModelInfoView } from "./ModelInfoView"
 import { DROPDOWN_Z_INDEX } from "./styles"
+import { RequestyModelPicker } from "./RequestyModelPicker"
 
 interface ApiOptionsProps {
 	apiErrorMessage?: string
@@ -154,6 +157,7 @@ const ApiOptions = ({ apiErrorMessage, modelIdErrorMessage }: ApiOptionsProps) =
 						{ value: "lmstudio", label: "LM Studio" },
 						{ value: "ollama", label: "Ollama" },
 						{ value: "unbound", label: "Unbound" },
+						{ value: "requesty", label: "Requesty" },
 					]}
 				/>
 			</div>
@@ -241,6 +245,27 @@ const ApiOptions = ({ apiErrorMessage, modelIdErrorMessage }: ApiOptionsProps) =
 				</div>
 			)}
 
+			{selectedProvider === "requesty" && (
+				<div>
+					<VSCodeTextField
+						value={apiConfiguration?.requestyApiKey || ""}
+						style={{ width: "100%" }}
+						type="password"
+						onInput={handleInputChange("requestyApiKey")}
+						placeholder="Enter API Key...">
+						<span style={{ fontWeight: 500 }}>Requesty API Key</span>
+					</VSCodeTextField>
+					<p
+						style={{
+							fontSize: "12px",
+							marginTop: "5px",
+							color: "var(--vscode-descriptionForeground)",
+						}}>
+						This key is stored locally and only used to make API requests from this extension.
+					</p>
+				</div>
+			)}
+
 			{selectedProvider === "openai-native" && (
 				<div>
 					<VSCodeTextField
@@ -1334,9 +1359,11 @@ const ApiOptions = ({ apiErrorMessage, modelIdErrorMessage }: ApiOptionsProps) =
 			{selectedProvider === "glama" && <GlamaModelPicker />}
 
 			{selectedProvider === "openrouter" && <OpenRouterModelPicker />}
+			{selectedProvider === "requesty" && <RequestyModelPicker />}
 
 			{selectedProvider !== "glama" &&
 				selectedProvider !== "openrouter" &&
+				selectedProvider !== "requesty" &&
 				selectedProvider !== "openai" &&
 				selectedProvider !== "ollama" &&
 				selectedProvider !== "lmstudio" &&
@@ -1478,6 +1505,12 @@ export function normalizeApiConfiguration(apiConfiguration?: ApiConfiguration) {
 				selectedModelId: apiConfiguration?.unboundModelId || unboundDefaultModelId,
 				selectedModelInfo: apiConfiguration?.unboundModelInfo || unboundDefaultModelInfo,
 			}
+		case "requesty":
+			return {
+				selectedProvider: provider,
+				selectedModelId: apiConfiguration?.requestyModelId || requestyDefaultModelId,
+				selectedModelInfo: apiConfiguration?.requestyModelInfo || requestyDefaultModelInfo,
+			}
 		default:
 			return getProviderData(anthropicModels, anthropicDefaultModelId)
 	}

webview-ui/src/components/settings/ModelPicker.tsx

@@ -25,10 +25,10 @@ import { ModelInfoView } from "./ModelInfoView"
 
 interface ModelPickerProps {
 	defaultModelId: string
-	modelsKey: "glamaModels" | "openRouterModels" | "unboundModels"
-	configKey: "glamaModelId" | "openRouterModelId" | "unboundModelId"
-	infoKey: "glamaModelInfo" | "openRouterModelInfo" | "unboundModelInfo"
-	refreshMessageType: "refreshGlamaModels" | "refreshOpenRouterModels" | "refreshUnboundModels"
+	modelsKey: "glamaModels" | "openRouterModels" | "unboundModels" | "requestyModels"
+	configKey: "glamaModelId" | "openRouterModelId" | "unboundModelId" | "requestyModelId"
+	infoKey: "glamaModelInfo" | "openRouterModelInfo" | "unboundModelInfo" | "requestyModelInfo"
+	refreshMessageType: "refreshGlamaModels" | "refreshOpenRouterModels" | "refreshUnboundModels" | "refreshRequestyModels"
 	serviceName: string
 	serviceUrl: string
 	recommendedModel: string

webview-ui/src/components/settings/RequestyModelPicker.tsx

@@ -0,0 +1,15 @@
+import { ModelPicker } from "./ModelPicker"
+import { requestyDefaultModelId } from "../../../../src/shared/api"
+
+export const RequestyModelPicker = () => (
+	<ModelPicker
+		defaultModelId={requestyDefaultModelId}
+		modelsKey="requestyModels"
+		configKey="requestyModelId"
+		infoKey="requestyModelInfo"
+		refreshMessageType="refreshRequestyModels"
+		serviceName="Requesty"
+		serviceUrl="https://requesty.ai"
+		recommendedModel="anthropic/claude-3-5-sonnet-latest"
+	/>
+)

webview-ui/src/context/ExtensionStateContext.tsx

@@ -10,6 +10,8 @@ import {
 	openRouterDefaultModelInfo,
 	unboundDefaultModelId,
 	unboundDefaultModelInfo,
+	requestyDefaultModelId,
+	requestyDefaultModelInfo,
 } from "../../../src/shared/api"
 import { vscode } from "../utils/vscode"
 import { convertTextMateToHljs } from "../utils/textMateToHljs"
@@ -25,6 +27,7 @@ export interface ExtensionStateContextType extends ExtensionState {
 	showWelcome: boolean
 	theme: any
 	glamaModels: Record<string, ModelInfo>
+	requestyModels: Record<string, ModelInfo>
 	openRouterModels: Record<string, ModelInfo>
 	unboundModels: Record<string, ModelInfo>
 	openAiModels: string[]
@@ -130,6 +133,9 @@ export const ExtensionStateContextProvider: React.FC<{ children: React.ReactNode
 	const [unboundModels, setUnboundModels] = useState<Record<string, ModelInfo>>({
 		[unboundDefaultModelId]: unboundDefaultModelInfo,
 	})
+	const [requestyModels, setRequestyModels] = useState<Record<string, ModelInfo>>({
+		[requestyDefaultModelId]: requestyDefaultModelInfo,
+	})
 
 	const [openAiModels, setOpenAiModels] = useState<string[]>([])
 	const [mcpServers, setMcpServers] = useState<McpServer[]>([])
@@ -250,6 +256,14 @@ export const ExtensionStateContextProvider: React.FC<{ children: React.ReactNode
 					setUnboundModels(updatedModels)
 					break
 				}
+				case "requestyModels": {
+					const updatedModels = message.requestyModels ?? {}
+					setRequestyModels({
+						[requestyDefaultModelId]: requestyDefaultModelInfo, // in case the extension sent a model list without the default model
+						...updatedModels,
+					})
+					break
+				}
 				case "mcpServers": {
 					setMcpServers(message.mcpServers ?? [])
 					break
@@ -279,6 +293,7 @@ export const ExtensionStateContextProvider: React.FC<{ children: React.ReactNode
 		showWelcome,
 		theme,
 		glamaModels,
+		requestyModels,
 		openRouterModels,
 		openAiModels,
 		unboundModels,