feat: image content

packages/ai/test/google-image-output.test.ts

@@ -0,0 +1,103 @@
+import { describe, expect, it, vi } from "vitest";
+
+vi.mock("@google/genai", () => {
+	class GoogleGenAI {
+		models = {
+			generateContentStream: async function* () {
+				yield {
+					responseId: "google-image-1",
+					candidates: [
+						{
+							content: {
+								parts: [
+									{ text: "Here is your image." },
+									{ inlineData: { mimeType: "image/png", data: "ZmFrZS1wbmc=" } },
+								],
+							},
+							finishReason: "STOP",
+						},
+					],
+					usageMetadata: {
+						promptTokenCount: 5,
+						candidatesTokenCount: 7,
+						totalTokenCount: 12,
+					},
+				};
+			},
+		};
+	}
+
+	return {
+		GoogleGenAI,
+		FinishReason: {
+			STOP: "STOP",
+			MAX_TOKENS: "MAX_TOKENS",
+			BLOCKLIST: "BLOCKLIST",
+			PROHIBITED_CONTENT: "PROHIBITED_CONTENT",
+			SPII: "SPII",
+			SAFETY: "SAFETY",
+			IMAGE_SAFETY: "IMAGE_SAFETY",
+			IMAGE_PROHIBITED_CONTENT: "IMAGE_PROHIBITED_CONTENT",
+			IMAGE_RECITATION: "IMAGE_RECITATION",
+			IMAGE_OTHER: "IMAGE_OTHER",
+			RECITATION: "RECITATION",
+			FINISH_REASON_UNSPECIFIED: "FINISH_REASON_UNSPECIFIED",
+			OTHER: "OTHER",
+			LANGUAGE: "LANGUAGE",
+			MALFORMED_FUNCTION_CALL: "MALFORMED_FUNCTION_CALL",
+			UNEXPECTED_TOOL_CALL: "UNEXPECTED_TOOL_CALL",
+			NO_IMAGE: "NO_IMAGE",
+		},
+		FunctionCallingConfigMode: {
+			AUTO: "AUTO",
+			ANY: "ANY",
+			NONE: "NONE",
+		},
+		ThinkingLevel: {
+			THINKING_LEVEL_UNSPECIFIED: "THINKING_LEVEL_UNSPECIFIED",
+			MINIMAL: "MINIMAL",
+			LOW: "LOW",
+			MEDIUM: "MEDIUM",
+			HIGH: "HIGH",
+		},
+	};
+});
+
+import { streamGoogle } from "../src/providers/google.js";
+import type { Context, Model } from "../src/types.js";
+
+describe("google image output", () => {
+	it("emits assistant image blocks for inlineData parts", async () => {
+		const model: Model<"google-generative-ai"> = {
+			id: "gemini-2.5-flash",
+			name: "Gemini 2.5 Flash",
+			api: "google-generative-ai",
+			provider: "google",
+			baseUrl: "https://generativelanguage.googleapis.com",
+			reasoning: false,
+			input: ["text", "image"],
+			cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
+			contextWindow: 1000000,
+			maxTokens: 8192,
+		};
+		const context: Context = {
+			messages: [{ role: "user", content: "Generate an image", timestamp: Date.now() }],
+		};
+
+		const result = streamGoogle(model, context, { apiKey: "test" });
+		const eventTypes: string[] = [];
+		for await (const event of result) {
+			eventTypes.push(event.type);
+		}
+
+		const message = await result.result();
+		expect(eventTypes).toEqual(["start", "text_start", "text_delta", "text_end", "image_start", "image_end", "done"]);
+		expect(message.responseId).toBe("google-image-1");
+		expect(message.content[0]).toMatchObject({ type: "text", text: "Here is your image." });
+		expect(message.content[1]).toMatchObject({
+			type: "image",
+			mimeType: "image/png",
+			data: "ZmFrZS1wbmc=",
+		});
+	});
+});

packages/ai/test/openai-completions-image-output.test.ts

@@ -0,0 +1,98 @@
+import { beforeEach, describe, expect, it, vi } from "vitest";
+import { streamSimple } from "../src/stream.js";
+import type { Context, Model } from "../src/types.js";
+
+const mockState = vi.hoisted(() => ({
+	lastParams: undefined as unknown,
+}));
+
+vi.mock("openai", () => {
+	class FakeOpenAI {
+		chat = {
+			completions: {
+				create: (params: unknown) => {
+					mockState.lastParams = params;
+					const response = {
+						id: "chatcmpl-image-1",
+						usage: {
+							prompt_tokens: 12,
+							completion_tokens: 34,
+							prompt_tokens_details: { cached_tokens: 0 },
+						},
+						choices: [
+							{
+								finish_reason: "stop",
+								message: {
+									role: "assistant",
+									content: "Here is your image.",
+									images: [
+										{
+											image_url: "data:image/png;base64,ZmFrZS1wbmc=",
+										},
+									],
+								},
+							},
+						],
+					};
+					const promise = Promise.resolve(response) as Promise<typeof response> & {
+						withResponse: () => Promise<{
+							data: typeof response;
+							response: { status: number; headers: Headers };
+						}>;
+					};
+					promise.withResponse = async () => ({
+						data: response,
+						response: { status: 200, headers: new Headers() },
+					});
+					return promise;
+				},
+			},
+		};
+	}
+
+	return { default: FakeOpenAI };
+});
+
+describe("openai-completions image output", () => {
+	beforeEach(() => {
+		mockState.lastParams = undefined;
+	});
+
+	it("switches OpenRouter image generation models to non-streaming and emits image events", async () => {
+		const model: Model<"openai-completions"> = {
+			id: "black-forest-labs/flux.2-pro",
+			name: "Black Forest Labs: FLUX.2 Pro",
+			api: "openai-completions",
+			provider: "openrouter",
+			baseUrl: "https://openrouter.ai/api/v1",
+			reasoning: false,
+			input: ["text", "image"],
+			cost: { input: 0.015, output: 0.03, cacheRead: 0, cacheWrite: 0 },
+			contextWindow: 100000,
+			maxTokens: 100000,
+		};
+		const context: Context = {
+			messages: [{ role: "user", content: "Generate a cat", timestamp: Date.now() }],
+		};
+
+		const result = streamSimple(model, context, { apiKey: "test" });
+		const eventTypes: string[] = [];
+		for await (const event of result) {
+			eventTypes.push(event.type);
+		}
+
+		const message = await result.result();
+		expect(eventTypes).toEqual(["start", "text_start", "text_delta", "text_end", "image_start", "image_end", "done"]);
+		expect(message.responseId).toBe("chatcmpl-image-1");
+		expect(message.content[0]).toMatchObject({ type: "text", text: "Here is your image." });
+		expect(message.content[1]).toMatchObject({
+			type: "image",
+			mimeType: "image/png",
+			data: "ZmFrZS1wbmc=",
+		});
+
+		const params = mockState.lastParams as { stream?: boolean; stream_options?: unknown };
+		expect(params.stream).toBe(false);
+		expect("stream_options" in (params as object)).toBe(false);
+	});
+});

packages/ai/test/openai-completions-thinking-as-text.test.ts

@@ -34,6 +34,7 @@ const compat = {
 	requiresReasoningContentOnAssistantMessages: false,
 	thinkingFormat: "openai",
 	openRouterRouting: {},
+	openRouterImageGeneration: false,
 	vercelGatewayRouting: {},
 	zaiToolStream: false,
 	supportsStrictMode: true,

packages/ai/test/openai-completions-tool-result-images.test.ts

@@ -32,6 +32,7 @@ const compat: Required<OpenAICompletionsCompat> = {
 	requiresReasoningContentOnAssistantMessages: false,
 	thinkingFormat: "openai",
 	openRouterRouting: {},
+	openRouterImageGeneration: false,
 	vercelGatewayRouting: {},
 	zaiToolStream: false,
 	supportsStrictMode: true,