20+ New AI Video Models Added: Wan 2.7, Seedance 2.0, Gemini Omni and More

We have given the SamAutomation AI catalog its biggest refresh yet. It's our biggest catalog yet — dozens of video and image models — and you reach every one of them through the same API key and the same unified endpoint. No new SDK, no per-vendor onboarding, no separate billing.

This post is a tour of what is new, grouped by what you are actually trying to do. If you just want to start building, jump to the AI API docs or query GET /api/ai/models for the live catalog with per-model credits and constraints.

One endpoint for every model

The whole catalog runs through a single job lifecycle, so adding a new model to your pipeline is a one-line change — swap the model_id:

1. POST /api/ai/jobs with a model_id, a prompt, and per-model options (resolution, duration, aspect ratio, audio). Add image_urls, video_urls or audio_urls for the models that take them.
2. You get back a job_id. Poll GET /api/ai/jobs/{id} until the status is success (with result_urls) or failed (credits are auto-refunded).
3. Credits scale with seconds × resolution, and GET /api/ai/models tells you the exact cost and accepted options for any model before you call it.

What is new, by category

Text to video

The core generation tier got deeper. Highlights:

• Wan 2.7 (Alibaba Tongyi) — a director-grade suite with text, image, reference-to-video and instruction-based editing, first/last-frame control and native audio.
• Kling 3.0 (Kuaishou), including a faster, lower-cost Turbo tier — multi-shot storyboarding with native audio.
• Veo 3.1 (Google DeepMind), now with a cost-efficient Lite tier alongside Fast and Quality — cinematic generation with synchronized native audio.

Multimodal & reference-driven

Steer generation with more than a prompt — images, video, voice or character references:

• Seedance 2.0 and Seedance 2.0 Fast (ByteDance) — combine text, image, video and audio references in a single call, with native in-pass audio and multi-shot cuts.
• Gemini Omni (Google) — any-input-to-video with voice and character-consistency workflows for personalized presenters.
• Kling 3.0 Motion Control — transfer the motion from a reference video onto your own character while keeping its identity.

Avatar & lip-sync

Two different paradigms, both now available:

• OmniHuman 1.5 (ByteDance) — animate a single still into an expressive talking avatar driven by a speech track, with context-aware gesture and head motion.
• Video Lip Sync — take existing talking-head footage and re-sync the mouth region to new target audio. Ideal for dubbing and localization without a reshoot.

Video editing

• Wan 2.7 and HappyHorse 1.1 (Alibaba) — instruction-based edits, reference-to-video and extend operations on existing clips. HappyHorse adds native multilingual lip-sync.

Pricing stayed simple

Everything is priced in credits, and credits scale with the work: a short 720p clip costs less than a long 1080p one on the same model. Your monthly plan credits cover the catalog, and you can top up at any time. Because pricing is computed from the live model metadata, the number you see in GET /api/ai/models is always the number you pay. Compare plans on the pricing page.

How to pick a model

The catalog is data-driven, so it stays current automatically — new models appear in GET /api/ai/models as we add them. Open the AI API docs to see request examples for each category, or read the JSON to Video API guide if you are combining AI generation with deterministic template rendering.