Veo 3.1

Veo 3.1 Lite's 4K-supported model.

Kling Motion Control

Kling AI 2.6 Motion Control API is a performance-driven image-to-video model. It can accurately transfer the real-person movements, postures, and expressions from the reference video to the character image, generating video frames with stable timing and realistic effects.

Gemini Omni Flash 1.1 Lite

Gemini Omni Flash 1.1 Lite fuses 1-7 reference images of characters, products or scenes into a consistent 10-second 720P or 1080P video with synchronized audio.

Muse Video

The Muse Video reference-to-video model supports native simultaneous audio generation (character lines and ambient sounds). It allows uploading up to 4 reference images (characters/objects/scenes) for fusion generation. It only supports 720P resolution. Available durations include 5/10/20/30 seconds or 40/50/60 seconds. Supported aspect ratios are landscape 16:9, portrait 9:16, square 1:1, 4:3 and 3:4. The actual duration of the final video may differ slightly from the selected duration (occasionally shorter or longer).

MiniMax H3 Lite

Minimax H3 low-cost channel reference video generation supports any combination of reference images (1~9), reference videos (up to 3), and reference audios (up to 3).

Kling 4.0

Kling 4.0 Reference Video Generation creates 3–30 second videos in 720p or 1080p using up to 10 reference images (@imageN), 5 reference videos (@videoN), 7 subjects and 3 timbres (all referenced via @names). A subject is made of 2–4 images or 1 short character clip to maintain consistency of characters, props or scenes, and character subjects can also be bound to their own voices. 🔸Note🔸: Video input has a lower per-second rate because the input duration is billed as well. No video input (text, images, image subjects or voices only): rate × output duration With video input: rate × (output × 2) when total input duration ≤ output duration; rate × (input + output) when input is longer. Multiple input videos are summed. Example: 10 s output + 5 s input = 20 s; with a 15 s input, 25 s.

Kling 3.0

Kling 3.0 Image-to-Video version supports multi-shot storytelling and can generate native audio, with a maximum controllable duration of 15 seconds. ‼️ When multishots is false: if the imageurls array length is 2, index 0 represents the first frame and index 1 represents the last frame; if the array length is 1, the array element serves as the first frame. ‼️ When multi_shots is true: only the first frame can be set. ‼️ Note that video generation with audio comes at a higher cost.

Wan 3.0 Video Prime Spicy

Wan 3.0 Prime Spicy I2V on APIPod animates a first frame (optionally with a last frame) into a 2 to 30 second clip with native audio at 480P, 720P, or 1080P on the faster tier of a relaxed-moderation upstream.