The focus of this release is to make production video generation faster, easier to monitor, and more flexible. Whether you are creating product videos, social ads, explainers, story-driven clips, or long-form image-list videos, the agent now gives you more control over render speed, model choice, voice style, and cost. Create videos in 1-shot upto 3 minutes long from text prompt or image list with the Samsar Vidgenie UI or API, with full post-processing capabilities in Studio.
You can now additionally also add narrator avatars to the videos, optional outro image with scannable CTA QR code and optional footer section with scannable CTA as well for marketing videos.
Faster renders and shorter queues
Long-running video agents are only useful when they can fit into real production workflows. This update reduces total queue duration significantly across common render sizes.
For many 1-minute videos, total queue time is now under 15 minutes. For 3-minute renders, total queue time is often under 30 minutes. Exact timing still depends on the selected video model, inference load, duration, and queue conditions, but average end-to-end wait time has come down substantially.
That means faster iteration, quicker client previews, and less waiting between prompt, review, and final export.
Live previews while rendering
You no longer need to wait until the full job completes to understand where the render is headed.
The agent now supports live previews during rendering, including while the job is still in queue. You can watch progress as the video comes together, making it easier to spot whether the visual direction, pacing, narration, and scene flow are matching your intent.
This is especially useful for longer videos, where waiting until the end to review every scene can slow down iteration. Live previews are the foundation for upcoming controls such as rollback, scene-level retry, and more granular edit recovery.
More model choice in Agent mode
The agent now supports a broader range of video models, so you can choose based on the tradeoff that matters most for the job: quality, speed, motion style, or cost.
Use lower-cost express models for fast production runs, higher-tier models when visual quality matters most, and custom Image-to-Video models when you want to bring your own infrastructure or optimize pricing.
Happy Horse 1.0 is now available
Happy Horse 1.0 is now available in Agent mode.
Happy Horse 1.0 is a state-of-the-art Image-to-Video model designed for strong motion, expressive image animation, and high-quality scene generation. It is now the default video model setting for Vidgenie UI and available alongside Runway, Seedance, Kling, and Veo options in supported agent workflows.
For image-list videos, this gives creators another strong option when starting from curated visuals, product images, generated frames, or storyboarded scenes.
Lower-cost custom Image-to-Video workflows
Now available for enterprise customers who want to generate videos at scale. If you use our default pipeline you are bottlenecked by the public queue for each stage.
Enterprises at scale can bring down the pricing significantly using their own API keys or Fal API compatible proxy server for image to video and text to video inference. Contact at hello@samsar.one for more information. If you use your own custom Image-to-Video model, you can bring video generation costs down dramatically.
Instead of using the highest-tier model pricing, custom Image-to-Video workflows can start as low as 14 credits per finished second, depending on account setup and adapter configuration. This makes longer videos more practical for teams generating at scale.
More TTS options in account settings
Voice selection is now more flexible too.
You can choose from OpenAI and ElevenLabs TTS speakers in account settings, making it easier to match narration to the tone of each video. Use a clear instructional voice for product walkthroughs, a warmer voice for storytelling, or a more energetic voice for ads and social content.
Updated cost comparison chart
Pricing below is shown in credits per finished video second. Optional add-ons such as narrator avatar generation and CTA generation may add to the final cost.
| Model | Workflow | Credits/sec | 1 min | 3 min |
|---|---|---|---|---|
| Runway Gen-4 | T2V + image-list | 30 | 1,800 | 5,400 |
| Seedance 1.5 | T2V + image-list | 30 | 1,800 | 5,400 |
| Veo 3.1 I2V Fast | T2V + image-list | 36 | 2,160 | 6,480 |
| Kling 3 Pro I2V | T2V + image-list | 36 | 2,160 | 6,480 |
| Kling 2.5 Turbo I2V | Image-list | 36 | 2,160 | 6,480 |
| Happy Horse 1.0 I2V | T2V + image-list | 36 | 2,160 | 6,480 |
| Veo 3.1 I2V | T2V + image-list | 60 | 3,600 | 10,800 |
Optional add-ons: narrator avatar generation adds 4 credits/sec, and express CTA generation adds 1 credit/sec.
Start building
The updated Samsar the One Text/Image-List-to-Video agent is now faster, more transparent, and more flexible. Try live previews, test Happy Horse 1.0, compare model pricing, and choose the right rendering path for your next video.
Optional add-ons: narrator avatar generation adds 4 credits/sec, and express CTA generation adds 1 credit/sec.
For integrations direct into enterprise workflows contact at hello@samsar.one