Add Vercel AI Gateway support

packages/ai/CHANGELOG.md

@@ -2,6 +2,10 @@
 
 ## [Unreleased]
 
+### Added
+
+- Added Vercel AI Gateway provider with model discovery and `AI_GATEWAY_API_KEY` env support.
+
 ## [0.45.3] - 2026-01-13
 
 ## [0.45.2] - 2026-01-13

packages/ai/README.md

@@ -56,6 +56,7 @@ Unified LLM API with automatic model discovery, provider configuration, token an
 - **Cerebras**
 - **xAI**
 - **OpenRouter**
+- **Vercel AI Gateway**
 - **MiniMax**
 - **GitHub Copilot** (requires OAuth, see below)
 - **Google Gemini CLI** (requires OAuth, see below)
@@ -862,6 +863,7 @@ In Node.js environments, you can set environment variables to avoid passing API
 | Cerebras | `CEREBRAS_API_KEY` |
 | xAI | `XAI_API_KEY` |
 | OpenRouter | `OPENROUTER_API_KEY` |
+| Vercel AI Gateway | `AI_GATEWAY_API_KEY` |
 | zAI | `ZAI_API_KEY` |
 | MiniMax | `MINIMAX_API_KEY` |
 | GitHub Copilot | `COPILOT_GITHUB_TOKEN` or `GH_TOKEN` or `GITHUB_TOKEN` |

packages/ai/scripts/generate-models.ts

@@ -32,6 +32,20 @@ interface ModelsDevModel {
 	};
 }
 
+interface AiGatewayModel {
+	id: string;
+	name?: string;
+	context_window?: number;
+	max_tokens?: number;
+	tags?: string[];
+	pricing?: {
+		input?: string | number;
+		output?: string | number;
+		input_cache_read?: string | number;
+		input_cache_write?: string | number;
+	};
+}
+
 const COPILOT_STATIC_HEADERS = {
 	"User-Agent": "GitHubCopilotChat/0.35.0",
 	"Editor-Version": "vscode/1.107.0",
@@ -39,6 +53,8 @@ const COPILOT_STATIC_HEADERS = {
 	"Copilot-Integration-Id": "vscode-chat",
 } as const;
 
+const AI_GATEWAY_BASE_URL = "https://ai-gateway.vercel.sh/v1";
+
 async function fetchOpenRouterModels(): Promise<Model<any>[]> {
 	try {
 		console.log("Fetching models from OpenRouter API...");
@@ -97,6 +113,64 @@ async function fetchOpenRouterModels(): Promise<Model<any>[]> {
 	}
 }
 
+async function fetchAiGatewayModels(): Promise<Model<any>[]> {
+	try {
+		console.log("Fetching models from Vercel AI Gateway API...");
+		const response = await fetch(`${AI_GATEWAY_BASE_URL}/models`);
+		const data = await response.json();
+		const models: Model<any>[] = [];
+
+		const toNumber = (value: string | number | undefined): number => {
+			if (typeof value === "number") {
+				return Number.isFinite(value) ? value : 0;
+			}
+			const parsed = parseFloat(value ?? "0");
+			return Number.isFinite(parsed) ? parsed : 0;
+		};
+
+		const items = Array.isArray(data.data) ? (data.data as AiGatewayModel[]) : [];
+		for (const model of items) {
+			const tags = Array.isArray(model.tags) ? model.tags : [];
+			// Only include models that support tools
+			if (!tags.includes("tool-use")) continue;
+
+			const input: ("text" | "image")[] = ["text"];
+			if (tags.includes("vision")) {
+				input.push("image");
+			}
+
+			const inputCost = toNumber(model.pricing?.input) * 1_000_000;
+			const outputCost = toNumber(model.pricing?.output) * 1_000_000;
+			const cacheReadCost = toNumber(model.pricing?.input_cache_read) * 1_000_000;
+			const cacheWriteCost = toNumber(model.pricing?.input_cache_write) * 1_000_000;
+
+			models.push({
+				id: model.id,
+				name: model.name || model.id,
+				api: "openai-completions",
+				baseUrl: AI_GATEWAY_BASE_URL,
+				provider: "ai-gateway",
+				reasoning: tags.includes("reasoning"),
+				input,
+				cost: {
+					input: inputCost,
+					output: outputCost,
+					cacheRead: cacheReadCost,
+					cacheWrite: cacheWriteCost,
+				},
+				contextWindow: model.context_window || 4096,
+				maxTokens: model.max_tokens || 4096,
+			});
+		}
+
+		console.log(`Fetched ${models.length} tool-capable models from Vercel AI Gateway`);
+		return models;
+	} catch (error) {
+		console.error("Failed to fetch Vercel AI Gateway models:", error);
+		return [];
+	}
+}
+
 async function loadModelsDevData(): Promise<Model<any>[]> {
 	try {
 		console.log("Fetching models from models.dev API...");
@@ -529,11 +603,13 @@ async function generateModels() {
 	// Fetch models from both sources
 	// models.dev: Anthropic, Google, OpenAI, Groq, Cerebras
 	// OpenRouter: xAI and other providers (excluding Anthropic, Google, OpenAI)
+	// AI Gateway: OpenAI-compatible catalog with tool-capable models
 	const modelsDevModels = await loadModelsDevData();
 	const openRouterModels = await fetchOpenRouterModels();
+	const aiGatewayModels = await fetchAiGatewayModels();
 
 	// Combine models (models.dev has priority)
-	const allModels = [...modelsDevModels, ...openRouterModels];
+	const allModels = [...modelsDevModels, ...openRouterModels, ...aiGatewayModels];
 
 	// Fix incorrect cache pricing for Claude Opus 4.5 from models.dev
 	// models.dev has 3x the correct pricing (1.5/18.75 instead of 0.5/6.25)

packages/ai/src/stream.ts

@@ -96,6 +96,7 @@ export function getEnvApiKey(provider: any): string | undefined {
 		cerebras: "CEREBRAS_API_KEY",
 		xai: "XAI_API_KEY",
 		openrouter: "OPENROUTER_API_KEY",
+		"ai-gateway": "AI_GATEWAY_API_KEY",
 		zai: "ZAI_API_KEY",
 		mistral: "MISTRAL_API_KEY",
 		minimax: "MINIMAX_API_KEY",

packages/ai/src/types.ts

@@ -56,6 +56,7 @@ export type KnownProvider =
 	| "groq"
 	| "cerebras"
 	| "openrouter"
+	| "ai-gateway"
 	| "zai"
 	| "mistral"
 	| "minimax"

packages/ai/test/stream.test.ts

@@ -598,6 +598,25 @@ describe("Generate E2E Tests", () => {
 		});
 	});
 
+	describe.skipIf(!process.env.AI_GATEWAY_API_KEY)(
+		"Vercel AI Gateway Provider (google/gemini-2.5-flash via OpenAI Completions)",
+		() => {
+			const llm = getModel("ai-gateway", "google/gemini-2.5-flash");
+
+			it("should complete basic text generation", { retry: 3 }, async () => {
+				await basicTextGeneration(llm);
+			});
+
+			it("should handle tool calling", { retry: 3 }, async () => {
+				await handleToolCall(llm);
+			});
+
+			it("should handle streaming", { retry: 3 }, async () => {
+				await handleStreaming(llm);
+			});
+		},
+	);
+
 	describe.skipIf(!process.env.ZAI_API_KEY)("zAI Provider (glm-4.5-air via OpenAI Completions)", () => {
 		const llm = getModel("zai", "glm-4.5-air");
 

packages/coding-agent/README.md

@@ -166,6 +166,7 @@ Add API keys to `~/.pi/agent/auth.json`:
 | Cerebras | `cerebras` | `CEREBRAS_API_KEY` |
 | xAI | `xai` | `XAI_API_KEY` |
 | OpenRouter | `openrouter` | `OPENROUTER_API_KEY` |
+| Vercel AI Gateway | `ai-gateway` | `AI_GATEWAY_API_KEY` |
 | ZAI | `zai` | `ZAI_API_KEY` |
 | MiniMax | `minimax` | `MINIMAX_API_KEY` |
 
@@ -1143,7 +1144,7 @@ pi [options] [@files...] [messages...]
 
 | Option | Description |
 |--------|-------------|
-| `--provider <name>` | Provider: `anthropic`, `openai`, `openai-codex`, `google`, `google-vertex`, `amazon-bedrock`, `mistral`, `xai`, `groq`, `cerebras`, `openrouter`, `zai`, `minimax`, `github-copilot`, `google-gemini-cli`, `google-antigravity`, or custom |
+| `--provider <name>` | Provider: `anthropic`, `openai`, `openai-codex`, `google`, `google-vertex`, `amazon-bedrock`, `mistral`, `xai`, `groq`, `cerebras`, `openrouter`, `ai-gateway`, `zai`, `minimax`, `github-copilot`, `google-gemini-cli`, `google-antigravity`, or custom |
 | `--model <id>` | Model ID |
 | `--api-key <key>` | API key (overrides environment) |
 | `--system-prompt <text\|file>` | Custom system prompt (text or file path) |

packages/coding-agent/src/cli/args.ts

@@ -242,6 +242,7 @@ ${chalk.bold("Environment Variables:")}
   CEREBRAS_API_KEY        - Cerebras API key
   XAI_API_KEY             - xAI Grok API key
   OPENROUTER_API_KEY      - OpenRouter API key
+  AI_GATEWAY_API_KEY      - Vercel AI Gateway API key
   ZAI_API_KEY             - ZAI API key
   MISTRAL_API_KEY         - Mistral API key
   MINIMAX_API_KEY         - MiniMax API key

packages/coding-agent/src/core/model-resolver.ts

@@ -21,6 +21,7 @@ export const defaultModelPerProvider: Record<KnownProvider, string> = {
 	"google-vertex": "gemini-3-pro-preview",
 	"github-copilot": "gpt-4o",
 	openrouter: "openai/gpt-5.1-codex",
+	"ai-gateway": "anthropic/claude-opus-4.5",
 	xai: "grok-4-fast-non-reasoning",
 	groq: "openai/gpt-oss-120b",
 	cerebras: "zai-glm-4.6",

packages/coding-agent/test/model-resolver.test.ts

@@ -1,6 +1,6 @@
 import type { Model } from "@mariozechner/pi-ai";
 import { describe, expect, test } from "vitest";
-import { parseModelPattern } from "../src/core/model-resolver.js";
+import { defaultModelPerProvider, findInitialModel, parseModelPattern } from "../src/core/model-resolver.js";
 
 // Mock models for testing
 const mockModels: Model<"anthropic-messages">[] = [
@@ -200,3 +200,37 @@ describe("parseModelPattern", () => {
 		});
 	});
 });
+
+describe("default model selection", () => {
+	test("ai-gateway default is opus 4.5", () => {
+		expect(defaultModelPerProvider["ai-gateway"]).toBe("anthropic/claude-opus-4.5");
+	});
+
+	test("findInitialModel selects ai-gateway default when available", async () => {
+		const aiGatewayModel: Model<"openai-completions"> = {
+			id: "anthropic/claude-opus-4.5",
+			name: "Claude Opus 4.5",
+			api: "openai-completions",
+			provider: "ai-gateway",
+			baseUrl: "https://ai-gateway.vercel.sh/v1",
+			reasoning: true,
+			input: ["text", "image"],
+			cost: { input: 5, output: 15, cacheRead: 0.5, cacheWrite: 5 },
+			contextWindow: 200000,
+			maxTokens: 8192,
+		};
+
+		const registry = {
+			getAvailable: async () => [aiGatewayModel],
+		} as unknown as Parameters<typeof findInitialModel>[0]["modelRegistry"];
+
+		const result = await findInitialModel({
+			scopedModels: [],
+			isContinuing: false,
+			modelRegistry: registry,
+		});
+
+		expect(result.model?.provider).toBe("ai-gateway");
+		expect(result.model?.id).toBe("anthropic/claude-opus-4.5");
+	});
+});

packages/web-ui/src/components/ProviderKeyInput.ts

@@ -15,6 +15,7 @@ const TEST_MODELS: Record<string, string> = {
 	google: "gemini-2.5-flash",
 	groq: "openai/gpt-oss-20b",
 	openrouter: "z-ai/glm-4.6",
+	"ai-gateway": "anthropic/claude-opus-4.5",
 	cerebras: "gpt-oss-120b",
 	xai: "grok-4-fast-non-reasoning",
 	zai: "glm-4.5-air",