@gitlab/gitlab-ai-provider 3.1.3 → 3.2.0

package/CHANGELOG.md

@@ -2,6 +2,10 @@
 
 All notable changes to this project will be documented in this file. See [Conventional Commits](https://conventionalcommits.org) for commit guidelines.
 
+## 3.2.0 (2026-01-22)
+
+- feat: add OpenAI models support with Responses API ([c715505](https://gitlab.com/gitlab-org/editor-extensions/gitlab-ai-provider/commit/c715505))
+
 ## <small>3.1.3 (2026-01-21)</small>
 
 - Merge branch 'fix/streaming-tool-call-handling' into 'main' ([22279b1](https://gitlab.com/gitlab-org/editor-extensions/gitlab-ai-provider/commit/22279b1))

package/README.md

@@ -4,7 +4,7 @@ A comprehensive TypeScript provider for integrating GitLab Duo AI capabilities w
 
 ## 🌟 Features
 
-- **🤖 Agentic Chat**: Native tool calling support via GitLab's Anthropic proxy
+- **🤖 Multi-Provider Agentic Chat**: Native tool calling support via GitLab's AI Gateway (Anthropic & OpenAI)
 - **🔐 Multiple Authentication**: Support for OAuth, Personal Access Tokens, and OpenCode auth
 - **🌐 Self-Hosted Support**: Works with both GitLab.com and self-hosted instances
 - **🔧 Tool Support**: Native tool calling via Vercel AI SDK
@@ -77,7 +77,7 @@ const { text } = await generateText({
 
 ### Model Variants
 
-The provider automatically maps specific model IDs to their corresponding Anthropic models:
+The provider automatically maps specific model IDs to their corresponding provider models (Anthropic or OpenAI) and routes requests to the appropriate AI Gateway proxy:
 
 ```typescript
 import { createGitLab } from '@gitlab/gitlab-ai-provider';
@@ -87,37 +87,110 @@ const gitlab = createGitLab({
   apiKey: process.env.GITLAB_TOKEN,
 });
 
-// Use Claude Opus 4.5
+// Anthropic Models (Claude)
 const opusModel = gitlab.agenticChat('duo-chat-opus-4-5');
 // Automatically uses: claude-opus-4-5-20251101
 
-// Use Claude Sonnet 4.5
 const sonnetModel = gitlab.agenticChat('duo-chat-sonnet-4-5');
 // Automatically uses: claude-sonnet-4-5-20250929
 
-// Use Claude Haiku 4.5
 const haikuModel = gitlab.agenticChat('duo-chat-haiku-4-5');
 // Automatically uses: claude-haiku-4-5-20251001
 
-// You can still override with explicit anthropicModel option
+// OpenAI Models (GPT-5)
+const gpt5Model = gitlab.agenticChat('duo-chat-gpt-5-1');
+// Automatically uses: gpt-5.1-2025-11-13
+
+const gpt5MiniModel = gitlab.agenticChat('duo-chat-gpt-5-mini');
+// Automatically uses: gpt-5-mini-2025-08-07
+
+const codexModel = gitlab.agenticChat('duo-chat-gpt-5-codex');
+// Automatically uses: gpt-5-codex
+
+// You can still override with explicit providerModel option
 const customModel = gitlab.agenticChat('duo-chat-opus-4-5', {
-  anthropicModel: 'claude-sonnet-4-5-20250929', // Override mapping
+  providerModel: 'claude-sonnet-4-5-20250929', // Override mapping
 });
 ```
 
 **Available Model Mappings:**
 
-| Model ID              | Anthropic Model              |
-| --------------------- | ---------------------------- |
-| `duo-chat-opus-4-5`   | `claude-opus-4-5-20251101`   |
-| `duo-chat-sonnet-4-5` | `claude-sonnet-4-5-20250929` |
-| `duo-chat-haiku-4-5`  | `claude-haiku-4-5-20251001`  |
+| Model ID                 | Provider  | Backend Model                |
+| ------------------------ | --------- | ---------------------------- |
+| `duo-chat-opus-4-5`      | Anthropic | `claude-opus-4-5-20251101`   |
+| `duo-chat-sonnet-4-5`    | Anthropic | `claude-sonnet-4-5-20250929` |
+| `duo-chat-haiku-4-5`     | Anthropic | `claude-haiku-4-5-20251001`  |
+| `duo-chat-gpt-5-1`       | OpenAI    | `gpt-5.1-2025-11-13`         |
+| `duo-chat-gpt-5-mini`    | OpenAI    | `gpt-5-mini-2025-08-07`      |
+| `duo-chat-gpt-5-codex`   | OpenAI    | `gpt-5-codex`                |
+| `duo-chat-gpt-5-2-codex` | OpenAI    | `gpt-5.2-codex`              |
+
+For unmapped Anthropic model IDs, the provider defaults to `claude-sonnet-4-5-20250929`.
+
+### OpenAI Models (GPT-5)
+
+The provider supports OpenAI GPT-5 models through GitLab's AI Gateway proxy. OpenAI models are automatically detected based on the model ID and routed to the appropriate proxy endpoint.
+
+```typescript
+import { createGitLab } from '@gitlab/gitlab-ai-provider';
+import { generateText } from 'ai';
+
+const gitlab = createGitLab({
+  apiKey: process.env.GITLAB_TOKEN,
+});
+
+// GPT-5.1 - Most capable model
+const { text } = await generateText({
+  model: gitlab.agenticChat('duo-chat-gpt-5-1'),
+  prompt: 'Explain GitLab CI/CD pipelines',
+});
+
+// GPT-5 Mini - Fast and efficient
+const { text: quickResponse } = await generateText({
+  model: gitlab.agenticChat('duo-chat-gpt-5-mini'),
+  prompt: 'Summarize this code',
+});
+
+// GPT-5 Codex - Optimized for code
+const { text: codeExplanation } = await generateText({
+  model: gitlab.agenticChat('duo-chat-gpt-5-codex'),
+  prompt: 'Refactor this function for better performance',
+});
+```
+
+**OpenAI Models with Tool Calling:**
+
+```typescript
+import { createGitLab } from '@gitlab/gitlab-ai-provider';
+import { generateText, tool } from 'ai';
+import { z } from 'zod';
+
+const gitlab = createGitLab({
+  apiKey: process.env.GITLAB_TOKEN,
+});
 
-For unmapped model IDs, the provider defaults to `claude-sonnet-4-5-20250929`.
+const { text, toolCalls } = await generateText({
+  model: gitlab.agenticChat('duo-chat-gpt-5-1', {
+    maxTokens: 4096,
+  }),
+  prompt: 'What is the weather in San Francisco?',
+  tools: {
+    getWeather: tool({
+      description: 'Get the weather for a location',
+      parameters: z.object({
+        location: z.string().describe('The city name'),
+      }),
+      execute: async ({ location }) => {
+        return { temperature: 72, condition: 'sunny', location };
+      },
+    }),
+  },
+});
+```
 
 ### Agentic Chat with Feature Flags
 
-You can pass feature flags to enable experimental features in GitLab's Anthropic proxy:
+You can pass feature flags to enable experimental features in GitLab's AI Gateway proxy:
 
 ```typescript
 import { createGitLab } from '@gitlab/gitlab-ai-provider';
@@ -215,7 +288,7 @@ interface GitLabProvider {
 }
 ```
 
-#### 2. **GitLabAgenticLanguageModel**
+#### 2. **GitLabAnthropicLanguageModel**
 
 Provides native tool calling through GitLab's Anthropic proxy.
 
@@ -224,6 +297,15 @@ Provides native tool calling through GitLab's Anthropic proxy.
 - Direct access token management
 - Supports all Anthropic tool calling features
 
+#### 3. **GitLabOpenAILanguageModel**
+
+Provides native tool calling through GitLab's OpenAI proxy.
+
+- Uses GPT-5 models via `https://cloud.gitlab.com/ai/v1/proxy/openai/`
+- Automatic token refresh and retry logic
+- Direct access token management
+- Supports all OpenAI tool calling features including parallel tool calls
+
 ### Supporting Utilities
 
 #### GitLabProjectDetector
@@ -309,11 +391,14 @@ interface GitLabProviderSettings {
 
 ```typescript
 interface GitLabAgenticOptions {
-  anthropicModel?: string; // Default: 'claude-sonnet-4-20250514'
+  providerModel?: string; // Override the backend model (e.g., 'claude-sonnet-4-5-20250929' or 'gpt-5.1-2025-11-13')
   maxTokens?: number; // Default: 8192
+  featureFlags?: Record<string, boolean>; // GitLab feature flags
 }
 ```
 
+**Note:** The `providerModel` option allows you to override the automatically mapped model. The provider will validate that the override is compatible with the model ID's provider (e.g., you cannot use an OpenAI model with a `duo-chat-opus-*` model ID).
+
 ### Error Handling
 
 ```typescript
@@ -365,19 +450,21 @@ npm run type-check    # TypeScript type checking
 ```
 gitlab-ai-provider/
 ├── src/
-│   ├── index.ts                          # Main exports
-│   ├── gitlab-provider.ts                # Provider factory
-│   ├── gitlab-agentic-language-model.ts  # Agentic chat model
-│   ├── gitlab-direct-access.ts           # Direct access tokens
-│   ├── gitlab-oauth-manager.ts           # OAuth management
-│   ├── gitlab-oauth-types.ts             # OAuth types
-│   ├── gitlab-project-detector.ts        # Project detection
-│   ├── gitlab-project-cache.ts           # Project caching
-│   ├── gitlab-api-types.ts               # API types
-│   ├── gitlab-error.ts                   # Error handling
-│   └── gitlab-workflow-debug.ts          # Debug logging
-├── tests/                                # Test files
-├── dist/                                 # Build output
+│   ├── index.ts                            # Main exports
+│   ├── gitlab-provider.ts                  # Provider factory
+│   ├── gitlab-anthropic-language-model.ts  # Anthropic/Claude model
+│   ├── gitlab-openai-language-model.ts     # OpenAI/GPT model
+│   ├── model-mappings.ts                   # Model ID mappings
+│   ├── gitlab-direct-access.ts             # Direct access tokens
+│   ├── gitlab-oauth-manager.ts             # OAuth management
+│   ├── gitlab-oauth-types.ts               # OAuth types
+│   ├── gitlab-project-detector.ts          # Project detection
+│   ├── gitlab-project-cache.ts             # Project caching
+│   ├── gitlab-api-types.ts                 # API types
+│   ├── gitlab-error.ts                     # Error handling
+│   └── gitlab-workflow-debug.ts            # Debug logging
+├── tests/                                  # Test files
+├── dist/                                   # Build output
 ├── package.json
 ├── tsconfig.json
 ├── tsup.config.ts
@@ -442,6 +529,7 @@ This project is built on top of:
 
 - [Vercel AI SDK](https://sdk.vercel.ai/)
 - [Anthropic SDK](https://github.com/anthropics/anthropic-sdk-typescript)
+- [OpenAI SDK](https://github.com/openai/openai-node)
 - [GitLab Duo](https://about.gitlab.com/gitlab-duo/)
 
 ---

package/dist/index.d.mts

@@ -1,7 +1,7 @@
 import { LanguageModelV2, LanguageModelV2CallOptions, LanguageModelV2Content, LanguageModelV2FinishReason, LanguageModelV2Usage, LanguageModelV2CallWarning, LanguageModelV2StreamPart } from '@ai-sdk/provider';
 import { z } from 'zod';
 
-interface GitLabAgenticConfig {
+interface GitLabAnthropicConfig {
     provider: string;
     instanceUrl: string;
     getHeaders: () => Record<string, string>;
@@ -36,20 +36,20 @@ interface GitLabAgenticConfig {
     aiGatewayUrl?: string;
 }
 /**
- * GitLab Agentic Language Model
+ * GitLab Anthropic Language Model
  *
  * This model uses GitLab's Anthropic proxy to provide native tool calling support
  * for the duo-chat model. It connects to Claude through GitLab's cloud proxy
  * at https://cloud.gitlab.com/ai/v1/proxy/anthropic/
  */
-declare class GitLabAgenticLanguageModel implements LanguageModelV2 {
+declare class GitLabAnthropicLanguageModel implements LanguageModelV2 {
     readonly specificationVersion: "v2";
     readonly modelId: string;
     readonly supportedUrls: Record<string, RegExp[]>;
     private readonly config;
     private readonly directAccessClient;
     private anthropicClient;
-    constructor(modelId: string, config: GitLabAgenticConfig);
+    constructor(modelId: string, config: GitLabAnthropicConfig);
     get provider(): string;
     /**
      * Get or create an Anthropic client with valid credentials
@@ -118,33 +118,33 @@ interface GitLabProvider {
      *   anthropicModel: 'claude-sonnet-4-5-20250929'
      * });
      */
-    agenticChat(modelId: string, options?: GitLabAgenticOptions): GitLabAgenticLanguageModel;
+    agenticChat(modelId: string, options?: GitLabAgenticOptions): GitLabAnthropicLanguageModel;
     textEmbeddingModel(modelId: string): never;
     imageModel(modelId: string): never;
 }
 interface GitLabAgenticOptions {
     /**
-     * The Anthropic model to use
+     * Override the provider-specific model (optional).
+     * Must be a valid model for the detected provider.
      *
-     * If not specified, automatically maps from the model ID:
-     * - 'duo-chat-opus-4-5' → 'claude-opus-4-5-20251101'
-     * - 'duo-chat-sonnet-4-5' → 'claude-sonnet-4-5-20250929'
-     * - 'duo-chat-haiku-4-5' → 'claude-haiku-4-5-20251001'
+     * For Anthropic models:
+     * - 'claude-opus-4-5-20251101'
+     * - 'claude-sonnet-4-5-20250929'
+     * - 'claude-haiku-4-5-20251001'
      *
-     * For unmapped model IDs, defaults to 'claude-sonnet-4-5-20250929'
-     *
-     * @default Automatically mapped from model ID, or 'claude-sonnet-4-5-20250929'
-     * @example
-     * // Use automatic mapping
-     * const model = gitlab.agenticChat('duo-chat-opus-4-5');
+     * For OpenAI models:
+     * - 'gpt-5.1-2025-11-13'
+     * - 'gpt-5-mini-2025-08-07'
+     * - 'gpt-5-codex'
+     * - 'gpt-5.2-codex'
      *
      * @example
      * // Override with explicit model
      * const model = gitlab.agenticChat('duo-chat-opus-4-5', {
-     *   anthropicModel: 'claude-sonnet-4-5-20250929'
+     *   providerModel: 'claude-sonnet-4-5-20250929'
      * });
      */
-    anthropicModel?: string;
+    providerModel?: string;
     /**
      * Maximum tokens to generate
      * @default 8192
@@ -214,32 +214,93 @@ declare function createGitLab(options?: GitLabProviderSettings): GitLabProvider;
  */
 declare const gitlab: GitLabProvider;
 
-/**
- * Maps GitLab model IDs to their corresponding Anthropic model identifiers.
- *
- * This mapping allows users to specify model variants by model ID without
- * needing to manually configure the anthropicModel option.
- *
- * @example
- * const model = gitlab.agenticChat('duo-chat-opus-4-5');
- * // Automatically uses 'claude-opus-4-5-20251101'
- */
-declare const MODEL_ID_TO_ANTHROPIC_MODEL: Record<string, string>;
-/**
- * Gets the Anthropic model identifier for a given GitLab model ID.
- *
- * @param modelId - The GitLab model ID (e.g., 'duo-chat-opus-4-5')
- * @returns The Anthropic model identifier, or undefined if no mapping exists
- *
- * @example
- * getAnthropicModelForModelId('duo-chat-opus-4-5')
- * // Returns: 'claude-opus-4-5-20251101'
- *
- * @example
- * getAnthropicModelForModelId('duo-chat')
- * // Returns: undefined (uses default)
- */
+interface GitLabOpenAIConfig {
+    provider: string;
+    instanceUrl: string;
+    getHeaders: () => Record<string, string>;
+    fetch?: typeof fetch;
+    refreshApiKey?: () => Promise<void>;
+    openaiModel?: string;
+    maxTokens?: number;
+    featureFlags?: {
+        DuoAgentPlatformNext: true;
+    } & Record<string, boolean>;
+    aiGatewayUrl?: string;
+    /** Whether to use the Responses API instead of Chat Completions API */
+    useResponsesApi?: boolean;
+}
+declare class GitLabOpenAILanguageModel implements LanguageModelV2 {
+    readonly specificationVersion: "v2";
+    readonly modelId: string;
+    readonly supportedUrls: Record<string, RegExp[]>;
+    private readonly config;
+    private readonly directAccessClient;
+    private readonly useResponsesApi;
+    private openaiClient;
+    constructor(modelId: string, config: GitLabOpenAIConfig);
+    get provider(): string;
+    private getOpenAIClient;
+    private isTokenError;
+    private convertTools;
+    private convertToolChoice;
+    private convertPrompt;
+    private convertFinishReason;
+    /**
+     * Convert tools to Responses API format
+     */
+    private convertToolsForResponses;
+    /**
+     * Convert prompt to Responses API input format
+     */
+    private convertPromptForResponses;
+    /**
+     * Extract system instructions from prompt
+     */
+    private extractSystemInstructions;
+    /**
+     * Convert Responses API status to finish reason
+     * Note: Responses API returns 'completed' even when making tool calls,
+     * so we need to check the content for tool calls separately.
+     */
+    private convertResponsesStatus;
+    doGenerate(options: LanguageModelV2CallOptions): Promise<{
+        content: LanguageModelV2Content[];
+        finishReason: LanguageModelV2FinishReason;
+        usage: LanguageModelV2Usage;
+        warnings: LanguageModelV2CallWarning[];
+    }>;
+    private doGenerateWithChatApi;
+    private doGenerateWithResponsesApi;
+    doStream(options: LanguageModelV2CallOptions): Promise<{
+        stream: ReadableStream<LanguageModelV2StreamPart>;
+        request?: {
+            body?: unknown;
+        };
+        response?: {
+            headers?: Record<string, string>;
+        };
+    }>;
+    private doStreamWithChatApi;
+    private doStreamWithResponsesApi;
+}
+
+type ModelProvider = 'anthropic' | 'openai';
+type OpenAIApiType = 'chat' | 'responses';
+interface ModelMapping {
+    provider: ModelProvider;
+    model: string;
+    /** For OpenAI models, which API to use: 'chat' for /v1/chat/completions, 'responses' for /v1/responses */
+    openaiApiType?: OpenAIApiType;
+}
+declare const MODEL_MAPPINGS: Record<string, ModelMapping>;
+declare function getModelMapping(modelId: string): ModelMapping | undefined;
+declare function getProviderForModelId(modelId: string): ModelProvider | undefined;
+declare function getValidModelsForProvider(provider: ModelProvider): string[];
 declare function getAnthropicModelForModelId(modelId: string): string | undefined;
+declare function getOpenAIModelForModelId(modelId: string): string | undefined;
+declare function getOpenAIApiType(modelId: string): OpenAIApiType;
+declare function isResponsesApiModel(modelId: string): boolean;
+declare const MODEL_ID_TO_ANTHROPIC_MODEL: Record<string, string>;
 
 interface GitLabErrorOptions {
     message: string;
@@ -265,15 +326,15 @@ declare const gitlabOAuthTokenResponseSchema: z.ZodObject<{
     expires_in: z.ZodNumber;
     created_at: z.ZodNumber;
 }, "strip", z.ZodTypeAny, {
+    created_at?: number;
     access_token?: string;
     refresh_token?: string;
     expires_in?: number;
-    created_at?: number;
 }, {
+    created_at?: number;
     access_token?: string;
     refresh_token?: string;
     expires_in?: number;
-    created_at?: number;
 }>;
 type GitLabOAuthTokenResponse = z.infer<typeof gitlabOAuthTokenResponseSchema>;
 
@@ -568,10 +629,16 @@ declare class GitLabDirectAccessClient {
      * Get the Anthropic proxy base URL
      */
     getAnthropicProxyUrl(): string;
+    /**
+     * Get the OpenAI proxy base URL
+     * Note: The OpenAI SDK expects a base URL like https://api.openai.com/v1
+     * and appends paths like /chat/completions. So we need /v1 at the end.
+     */
+    getOpenAIProxyUrl(): string;
     /**
      * Invalidate the cached token
      */
     invalidateToken(): void;
 }
 
-export { BUNDLED_CLIENT_ID, DEFAULT_AI_GATEWAY_URL, type DirectAccessToken, GITLAB_COM_URL, type GitLabAgenticConfig, GitLabAgenticLanguageModel, type GitLabAgenticOptions, GitLabDirectAccessClient, type GitLabDirectAccessConfig, GitLabError, type GitLabErrorOptions, GitLabOAuthManager, type GitLabOAuthTokenResponse, type GitLabOAuthTokens, type GitLabProject, GitLabProjectCache, GitLabProjectDetector, type GitLabProjectDetectorConfig, type GitLabProvider, type GitLabProviderSettings, MODEL_ID_TO_ANTHROPIC_MODEL, OAUTH_SCOPES, type OpenCodeAuth, type OpenCodeAuthApi, type OpenCodeAuthOAuth, TOKEN_EXPIRY_SKEW_MS, createGitLab, getAnthropicModelForModelId, gitlab };
+export { BUNDLED_CLIENT_ID, DEFAULT_AI_GATEWAY_URL, type DirectAccessToken, GITLAB_COM_URL, type GitLabAgenticOptions, type GitLabAnthropicConfig, GitLabAnthropicLanguageModel, GitLabDirectAccessClient, type GitLabDirectAccessConfig, GitLabError, type GitLabErrorOptions, GitLabOAuthManager, type GitLabOAuthTokenResponse, type GitLabOAuthTokens, type GitLabOpenAIConfig, GitLabOpenAILanguageModel, type GitLabProject, GitLabProjectCache, GitLabProjectDetector, type GitLabProjectDetectorConfig, type GitLabProvider, type GitLabProviderSettings, MODEL_ID_TO_ANTHROPIC_MODEL, MODEL_MAPPINGS, type ModelMapping, type ModelProvider, OAUTH_SCOPES, type OpenAIApiType, type OpenCodeAuth, type OpenCodeAuthApi, type OpenCodeAuthOAuth, TOKEN_EXPIRY_SKEW_MS, createGitLab, getAnthropicModelForModelId, getModelMapping, getOpenAIApiType, getOpenAIModelForModelId, getProviderForModelId, getValidModelsForProvider, gitlab, isResponsesApiModel };