fix: align video generation payloads

apps/api/internal/runner/param_processor_test.go

@@ -6,6 +6,50 @@ import (
 	"github.com/easyai/easyai-ai-gateway/apps/api/internal/store"
 )
 
+func TestVideoModelTypeInferenceReadsContentArray(t *testing.T) {
+	imageToVideo := modelTypeFromKind("videos.generations", map[string]any{
+		"model": "demo-video",
+		"content": []any{
+			map[string]any{"type": "text", "text": "animate it"},
+			map[string]any{"type": "image_url", "role": "first_frame", "image_url": map[string]any{"url": "https://example.com/frame.png"}},
+		},
+	})
+	if imageToVideo != "image_to_video" {
+		t.Fatalf("image content should infer image_to_video, got %s", imageToVideo)
+	}
+
+	omniVideo := modelTypeFromKind("videos.generations", map[string]any{
+		"model": "demo-video",
+		"content": []any{
+			map[string]any{"type": "text", "text": "edit it"},
+			map[string]any{"type": "video_url", "role": "reference_video", "video_url": map[string]any{"url": "https://example.com/ref.mp4"}},
+		},
+	})
+	if omniVideo != "omni_video" {
+		t.Fatalf("video content should infer omni_video, got %s", omniVideo)
+	}
+
+	textToVideo := modelTypeFromKind("videos.generations", map[string]any{
+		"model":   "demo-video",
+		"content": []any{map[string]any{"type": "text", "text": "make a clip"}},
+	})
+	if textToVideo != "video_generate" {
+		t.Fatalf("text-only content should infer video_generate, got %s", textToVideo)
+	}
+}
+
+func TestVideoContentTextContributesToTokenEstimate(t *testing.T) {
+	tokens := estimateRequestTokens(map[string]any{
+		"model": "demo-video",
+		"content": []any{
+			map[string]any{"type": "text", "text": "a cinematic product reveal"},
+		},
+	})
+	if tokens <= 1 {
+		t.Fatalf("content text should contribute to token estimate, got %d", tokens)
+	}
+}
+
 func TestParamProcessorOmniFiltersUnsupportedVideoAndAudioContent(t *testing.T) {
 	body := map[string]any{
 		"model":  "可灵O1",