feat: Add SAP AI Core provider integration

- Add comprehensive SAP AI Core provider
- Implement authentication using OAuth 2.0 client credentials flow
- Add deployment detection and model mapping for SAP AI Core services
- Include streaming support for real-time responses
- Add comprehensive test coverage for all provider functionality
- Implement UI components for SAP AI Core configuration and model selection
- Add model picker with deployment status indication
- Include validation for required configuration fields
- Add internationalization support for SAP AI Core settings
- Integrate with existing provider architecture and settings system

This enables users to connect to SAP AI Core's enterprise AI platform,
supporting both Azure OpenAI and GCP Vertex AI model deployments with
proper authentication and deployment management.

Aligns with roadmap goals:
- Enhanced AI provider support for enterprise platforms
- Reliable authentication and connection handling
- Improved user experience with clear deployment status

Author: junaidsap

packages/types/src/provider-settings.ts

@@ -20,6 +20,7 @@ import {
 	openAiNativeModels,
 	rooModels,
 	sambaNovaModels,
+	sapAiCoreModels,
 	vertexModels,
 	vscodeLlmModels,
 	xaiModels,
@@ -64,6 +65,7 @@ export const providerNames = [
 	"featherless",
 	"io-intelligence",
 	"roo",
+	"sapaicore",
 ] as const
 
 export const providerNamesSchema = z.enum(providerNames)
@@ -315,6 +317,16 @@ const rooSchema = apiModelIdProviderModelSchema.extend({
 	// No additional fields needed - uses cloud authentication
 })
 
+const sapAiCoreSchema = apiModelIdProviderModelSchema.extend({
+	sapAiCoreClientId: z.string().optional(),
+	sapAiCoreClientSecret: z.string().optional(),
+	sapAiCoreTokenUrl: z.string().optional(),
+	sapAiResourceGroup: z.string().optional(),
+	sapAiCoreBaseUrl: z.string().optional(),
+	reasoningEffort: z.string().optional(),
+	thinkingBudgetTokens: z.number().optional(),
+})
+
 const defaultSchema = z.object({
 	apiProvider: z.undefined(),
 })
@@ -353,6 +365,7 @@ export const providerSettingsSchemaDiscriminated = z.discriminatedUnion("apiProv
 	featherlessSchema.merge(z.object({ apiProvider: z.literal("featherless") })),
 	ioIntelligenceSchema.merge(z.object({ apiProvider: z.literal("io-intelligence") })),
 	rooSchema.merge(z.object({ apiProvider: z.literal("roo") })),
+	sapAiCoreSchema.merge(z.object({ apiProvider: z.literal("sapaicore") })),
 	defaultSchema,
 ])
 
@@ -391,6 +404,7 @@ export const providerSettingsSchema = z.object({
 	...featherlessSchema.shape,
 	...ioIntelligenceSchema.shape,
 	...rooSchema.shape,
+	...sapAiCoreSchema.shape,
 	...codebaseIndexProviderSchema.shape,
 })
 
@@ -512,6 +526,11 @@ export const MODELS_BY_PROVIDER: Record<
 		label: "SambaNova",
 		models: Object.keys(sambaNovaModels),
 	},
+	sapaicore: {
+		id: "sapaicore",
+		label: "SAP AI Core",
+		models: Object.keys(sapAiCoreModels),
+	},
 	vertex: {
 		id: "vertex",
 		label: "GCP Vertex AI",

packages/types/src/providers/index.ts

@@ -22,6 +22,7 @@ export * from "./openrouter.js"
 export * from "./requesty.js"
 export * from "./roo.js"
 export * from "./sambanova.js"
+export * from "./sapaicore.js"
 export * from "./unbound.js"
 export * from "./vertex.js"
 export * from "./vscode-llm.js"

packages/types/src/providers/sapaicore.ts

@@ -0,0 +1,187 @@
+import type { ModelInfo } from "../model.js"
+
+export type SapAiCoreModelId = keyof typeof sapAiCoreModels
+
+export const sapAiCoreDefaultModelId: SapAiCoreModelId = "anthropic--claude-3.5-sonnet"
+
+export const sapAiCoreModels = {
+	// Anthropic models
+	"anthropic--claude-4-sonnet": {
+		maxTokens: 8192,
+		contextWindow: 2000000,
+		supportsImages: true,
+		supportsPromptCache: true,
+		inputPrice: 5.0,
+		outputPrice: 15.0,
+		cacheReadsPrice: 0.5,
+		cacheWritesPrice: 7.5,
+	},
+	"anthropic--claude-4-opus": {
+		maxTokens: 8192,
+		contextWindow: 2000000,
+		supportsImages: true,
+		supportsPromptCache: true,
+		inputPrice: 20.0,
+		outputPrice: 60.0,
+		cacheReadsPrice: 2.0,
+		cacheWritesPrice: 30.0,
+	},
+	"anthropic--claude-3.7-sonnet": {
+		maxTokens: 8192,
+		contextWindow: 1000000,
+		supportsImages: true,
+		supportsPromptCache: true,
+		inputPrice: 3.0,
+		outputPrice: 15.0,
+		cacheReadsPrice: 0.3,
+		cacheWritesPrice: 4.5,
+	},
+	"anthropic--claude-3.5-sonnet": {
+		maxTokens: 8192,
+		contextWindow: 200000,
+		supportsImages: true,
+		supportsPromptCache: false,
+		inputPrice: 3.0,
+		outputPrice: 15.0,
+	},
+	"anthropic--claude-3-sonnet": {
+		maxTokens: 4096,
+		contextWindow: 200000,
+		supportsImages: true,
+		supportsPromptCache: false,
+		inputPrice: 3.0,
+		outputPrice: 15.0,
+	},
+	"anthropic--claude-3-haiku": {
+		maxTokens: 4096,
+		contextWindow: 200000,
+		supportsImages: true,
+		supportsPromptCache: false,
+		inputPrice: 0.25,
+		outputPrice: 1.25,
+	},
+	"anthropic--claude-3-opus": {
+		maxTokens: 4096,
+		contextWindow: 200000,
+		supportsImages: true,
+		supportsPromptCache: false,
+		inputPrice: 15.0,
+		outputPrice: 75.0,
+	},
+	// OpenAI models
+	"gpt-4o": {
+		maxTokens: 16384,
+		contextWindow: 128000,
+		supportsImages: true,
+		supportsPromptCache: false,
+		inputPrice: 5.0,
+		outputPrice: 15.0,
+	},
+	"gpt-4": {
+		maxTokens: 8192,
+		contextWindow: 8192,
+		supportsImages: false,
+		supportsPromptCache: false,
+		inputPrice: 30.0,
+		outputPrice: 60.0,
+	},
+	"gpt-4o-mini": {
+		maxTokens: 16384,
+		contextWindow: 128000,
+		supportsImages: true,
+		supportsPromptCache: false,
+		inputPrice: 0.15,
+		outputPrice: 0.6,
+	},
+	o1: {
+		maxTokens: 100000,
+		contextWindow: 200000,
+		supportsImages: false,
+		supportsPromptCache: false,
+		inputPrice: 15.0,
+		outputPrice: 60.0,
+	},
+	"gpt-4.1": {
+		maxTokens: 32768,
+		contextWindow: 128000,
+		supportsImages: true,
+		supportsPromptCache: false,
+		inputPrice: 5.0,
+		outputPrice: 15.0,
+	},
+	"gpt-4.1-nano": {
+		maxTokens: 16384,
+		contextWindow: 128000,
+		supportsImages: true,
+		supportsPromptCache: false,
+		inputPrice: 0.15,
+		outputPrice: 0.6,
+	},
+	"gpt-5": {
+		maxTokens: 32768,
+		contextWindow: 256000,
+		supportsImages: true,
+		supportsPromptCache: false,
+		inputPrice: 10.0,
+		outputPrice: 30.0,
+	},
+	"gpt-5-nano": {
+		maxTokens: 16384,
+		contextWindow: 256000,
+		supportsImages: true,
+		supportsPromptCache: false,
+		inputPrice: 0.5,
+		outputPrice: 1.5,
+	},
+	"gpt-5-mini": {
+		maxTokens: 16384,
+		contextWindow: 256000,
+		supportsImages: true,
+		supportsPromptCache: false,
+		inputPrice: 1.0,
+		outputPrice: 3.0,
+	},
+	"o3-mini": {
+		maxTokens: 100000,
+		contextWindow: 200000,
+		supportsImages: false,
+		supportsPromptCache: false,
+		inputPrice: 1.0,
+		outputPrice: 4.0,
+	},
+	o3: {
+		maxTokens: 100000,
+		contextWindow: 200000,
+		supportsImages: false,
+		supportsPromptCache: false,
+		inputPrice: 60.0,
+		outputPrice: 240.0,
+	},
+	"o4-mini": {
+		maxTokens: 100000,
+		contextWindow: 200000,
+		supportsImages: false,
+		supportsPromptCache: false,
+		inputPrice: 2.0,
+		outputPrice: 8.0,
+	},
+	// Gemini models
+	"gemini-2.5-flash": {
+		maxTokens: 8192,
+		contextWindow: 1000000,
+		supportsImages: true,
+		supportsPromptCache: false,
+		inputPrice: 0.5,
+		outputPrice: 1.5,
+		maxThinkingTokens: 32768,
+	},
+	"gemini-2.5-pro": {
+		maxTokens: 8192,
+		contextWindow: 2000000,
+		supportsImages: true,
+		supportsPromptCache: false,
+		inputPrice: 2.5,
+		outputPrice: 10.0,
+		maxThinkingTokens: 65536,
+	},
+} satisfies Record<string, ModelInfo>

src/api/index.ts

@@ -37,6 +37,7 @@ import {
 	FireworksHandler,
 	RooHandler,
 	FeatherlessHandler,
+	SapAiCoreHandler,
 } from "./providers"
 import { NativeOllamaHandler } from "./providers/native-ollama"
 
@@ -148,6 +149,8 @@ export function buildApiHandler(configuration: ProviderSettings): ApiHandler {
 			return new RooHandler(options)
 		case "featherless":
 			return new FeatherlessHandler(options)
+		case "sapaicore":
+			return new SapAiCoreHandler(options)
 		default:
 			apiProvider satisfies "gemini-cli" | undefined
 			return new AnthropicHandler(options)