feat: cost for OpenRouter image models

.changeset/openrouter-image-cost.md

@@ -0,0 +1,6 @@
+---
+'@tanstack/ai-openrouter': patch
+---
+
+Surface OpenRouter provider-reported cost on image generation usage, matching
+the existing text adapter behavior.

packages/ai-openrouter/src/adapters/image.ts

@@ -6,6 +6,7 @@ import {
   generateId as utilGenerateId,
 } from '../utils'
 import { buildOpenRouterUsage } from '../usage'
+import { extractUsageCost } from './cost'
 import type { OpenRouterClientConfig } from '../utils'
 import type {
   OpenRouterImageModelInputModalitiesByName,
@@ -203,10 +204,10 @@ export class OpenRouterImageAdapter<
 
     // OpenRouter routes image generation through the chat surface, so the
     // response carries the same `usage` shape as text. Surface it (with any
-    // detail breakdowns) when present.
-    const usage = response.usage
-      ? buildOpenRouterUsage(response.usage)
-      : undefined
+    // detail breakdowns and provider-reported cost) when present.
+    const baseUsage = buildOpenRouterUsage(response.usage)
+    const usage =
+      baseUsage && { ...baseUsage, ...extractUsageCost(response.usage) }
 
     return {
       id: response.id || this.generateId(),

packages/ai-openrouter/tests/image-adapter.test.ts

@@ -111,6 +111,45 @@ describe('OpenRouter Image Adapter', () => {
     })
   })
 
+  it('surfaces provider-reported cost from OpenRouter image usage', async () => {
+    const mockResponse = {
+      ...createMockImageResponse([{ url: 'https://example.com/image1.png' }]),
+      usage: {
+        completionTokens: 1291,
+        cost: 0.0387076,
+        cost_details: {
+          upstream_inference_completions_cost: 0.0387025,
+          upstream_inference_cost: 0.0387076,
+          upstream_inference_prompt_cost: 0.0000051,
+        },
+        promptTokens: 17,
+        totalTokens: 1308,
+      },
+    }
+
+    mockSend = vi.fn().mockResolvedValueOnce(mockResponse)
+
+    const adapter = createAdapter()
+
+    const result = await adapter.generateImages({
+      model: 'google/gemini-2.5-flash-image',
+      prompt: 'A futuristic city at sunset',
+      logger: testLogger,
+    })
+
+    expect(result.usage).toMatchObject({
+      promptTokens: 17,
+      completionTokens: 1291,
+      totalTokens: 1308,
+      cost: 0.0387076,
+      costDetails: {
+        upstreamOutputCost: 0.0387025,
+        upstreamCost: 0.0387076,
+        upstreamInputCost: 0.0000051,
+      },
+    })
+  })
+
   it('generates multiple images', async () => {
     const mockResponse = createMockImageResponse([
       { url: 'https://example.com/image1.png' },