Veo 3.1 Lite was released by Google on March 31, 2026 as a cost-effective video generation model (less than 50% cost of Veo 3.1 Fast with same speed). However, when attempting to use Veo 3.1 Lite model keys via FlowKit's video generation pipeline, the model is not recognized/found.
Expected Behavior
FlowKit should support Veo 3.1 Lite as a video model option, allowing users to configure it via the /api/flow/settings endpoint or model configuration. Expected model keys:
- veo_3_1_i2v_lite_portrait (9:16)
- veo_3_1_i2v_lite_landscape (16:9)
Current Behavior
When setting video models to Veo 3.1 Lite variants via the settings API (video_models.PAYGATE_TIER_TWO.frame_2_video), video generation fails with model not found or the request is rejected by the Google Flow API.
The current hardcoded VIDEO_MODELS in flow_client.py only includes Veo 3.1 (standard/fast) but not the Lite variant.
Steps to Reproduce
-
Configure FlowKit video model settings to use Veo 3.1 Lite:
{
"video_models": {
"PAYGATE_TIER_TWO": {
"frame_2_video": {
"VIDEO_ASPECT_RATIO_PORTRAIT": "veo_3_1_i2v_lite_portrait",
"VIDEO_ASPECT_RATIO_LANDSCAPE": "veo_3_1_i2v_lite_landscape"
},
"start_end_frame_2_video": {
"VIDEO_ASPECT_RATIO_PORTRAIT": "veo_3_1_i2v_lite_portrait",
"VIDEO_ASPECT_RATIO_LANDSCAPE": "veo_3_1_i2v_lite_landscape"
}
}
}
}
-
Trigger video generation via /fk-gen-videos or API
-
Generation fails — model not found
Context
- Veo 3.1 Lite is available via Gemini API and Google AI Studio (paid tier)
- It supports both text-to-video and image-to-video (i2v)
- Output: 720p/1080p, 9:16 and 16:9, 4/6/8 second clips
- Cost: <50% of Veo 3.1 Fast
- Speed: Same as Veo 3.1 Fast
Suggested Fix
Add Veo 3.1 Lite model keys to the VIDEO_MODELS dictionary and expose them as a selectable tier option (e.g., alongside existing TIER_ONE/TIER_TWO models), or allow the model key to be fully user-configurable without validation against a hardcoded list.
Environment
- FlowKit version: latest (main branch)
- Google account tier: PAYGATE_TIER_TWO
- OS: Windows 11 + Chrome extension
Veo 3.1 Lite was released by Google on March 31, 2026 as a cost-effective video generation model (less than 50% cost of Veo 3.1 Fast with same speed). However, when attempting to use Veo 3.1 Lite model keys via FlowKit's video generation pipeline, the model is not recognized/found.
Expected Behavior
FlowKit should support Veo 3.1 Lite as a video model option, allowing users to configure it via the /api/flow/settings endpoint or model configuration. Expected model keys:
Current Behavior
When setting video models to Veo 3.1 Lite variants via the settings API (video_models.PAYGATE_TIER_TWO.frame_2_video), video generation fails with model not found or the request is rejected by the Google Flow API.
The current hardcoded VIDEO_MODELS in flow_client.py only includes Veo 3.1 (standard/fast) but not the Lite variant.
Steps to Reproduce
Configure FlowKit video model settings to use Veo 3.1 Lite:
{
"video_models": {
"PAYGATE_TIER_TWO": {
"frame_2_video": {
"VIDEO_ASPECT_RATIO_PORTRAIT": "veo_3_1_i2v_lite_portrait",
"VIDEO_ASPECT_RATIO_LANDSCAPE": "veo_3_1_i2v_lite_landscape"
},
"start_end_frame_2_video": {
"VIDEO_ASPECT_RATIO_PORTRAIT": "veo_3_1_i2v_lite_portrait",
"VIDEO_ASPECT_RATIO_LANDSCAPE": "veo_3_1_i2v_lite_landscape"
}
}
}
}
Trigger video generation via /fk-gen-videos or API
Generation fails — model not found
Context
Suggested Fix
Add Veo 3.1 Lite model keys to the VIDEO_MODELS dictionary and expose them as a selectable tier option (e.g., alongside existing TIER_ONE/TIER_TWO models), or allow the model key to be fully user-configurable without validation against a hardcoded list.
Environment