video generation
AI Video Models 2027
The most complete 2027 lineup in one place: every video model Felo carries, from text-to-video to reference-driven routes, with live rates and one key.
- 2027 lineup
- All providers in one key
- Live per-second prices
14 models · 8 Providers
The 2027 lineup
14 models
2027 picks
TBC: Mischief 1NewTBC: Mischief 1 is a The Biological Computing Company model available through the OpenRouter API, priced from 0.005 US dollars per second of usage.Resolution: 480p / 720pDuration: 3–5s02
Kling 4.0 Image-to-VideoNewKling 4.0 Image-to-Video is a unified creative model that animates reference images into native 3–30 second videos with stereo audio, multimodal references, and built-in editing tools.Resolution: 720p / 1080pDuration: 3–30s03Kling 4.0 Text-to-VideoNewKling 4.0 Text-to-Video turns written prompts into 3–30 second videos with native stereo audio, supporting multi-shot storyboards and multilingual dialogue.Resolution: 720p / 1080pDuration: 3–30s04Kling 4.0 Omni-VideoNewKling 4.0 Omni is a reference-driven video generation model from Kuaishou that creates and edits video from up to 15 combined image, video, audio, and element references.Resolution: 720p / 1080pDuration: 3–30s05
xAI: Grok Imagine Video 1.5 LiteNewSpaceXAI's cost-efficient, distilled text- and image-to-video model with native synchronized audio, 1–15 second clips, and resolutions up to 1080p.$0.02/ 1 secondsPricingResolution: 480p / 720p / 1080pDuration: 1–15s06
HeyGen VideoNewHeyGen Video is an AI video generation model from HeyGen that creates scene visuals, dialogue and sound effects in a single pass, ranked #1 on the Artificial Analysis video leaderboard as of 2026.$0.02/ 1 secondsPricingResolution: 480p / 768p / 2KDuration: 5–15s07
MiniMax: H3 MaxNewMiniMax: H3 Max is a model listed under the name MiniMax: H3 Max, with no further details provided in the available materials.$0.08/ 1 secondsPricingResolution: 768p / 480pDuration: 5–15s08
Alibaba: Wan 3.0NewAlibaba Tongyi Lab's Wan 3.0 is a video generation model that produces up to 30-second native clips with synchronized audio from text, images, video, and documents.$0.05/ 1 secondsPricingResolution: 480p / 720p / 1080pDuration: 2–30s09
ByteDance: Seedance 2.0 MiniNewSeedance 2.0 Mini is ByteDance's compact, cost-efficient video generation model for high-volume short clips, with text, image, video and audio references.Resolution: 480p / 720pDuration: 4–15s10ByteDance: Seedance 2.5NewByteDance Seed's Seedance 2.5 generates up to 30-second continuous video with synchronized audio from text, images, video and audio references.$10.7/ 1M tokensPricingResolution: 480p / 720p / 1080pDuration: 4–30s11MiniMax: H3NewMiniMax H3 is a 33B-parameter open-weights diffusion transformer that unifies text, image, video, and audio understanding with native video and stereo audio generation.Resolution: 2KDuration: 5–15s12
Google: Veo 3.1 FastVeo 3.1 Fast is Google's premium first-party model for text-to-video and image-to-video generation, available in Google Cloud Model Garden.$0.08/ 1 secondsPricingResolution: 720p / 1080p / 4KDuration: 4–8s13ByteDance: Seedance 2.0ByteDance's Seedance 2.0 is a natively multimodal video foundation model that generates synchronized audio and video from text, image, clip, and audio inputs.$7/ 1M tokensPricingResolution: 480p / 720p / 1080p / 4KDuration: 4–15s14ByteDance: Seedance 2.0 FastByteDance's speed-optimized Seedance 2.0 Fast video model generates 4–15 second clips up to 1080p with synchronized audio from text, image, video, and audio inputs.$4.2/ 1M tokensPricingResolution: 480p / 720pDuration: 4–15sWhat you can build
Ad creative
Turn a product shot and a script into vertical and square spots, then re-run the same prompt at a different resolution.
Product demos
Render a short walkthrough from screenshots and a first and last frame, without a camera or an edit suite.
Social clips
Test hooks at volume: one prompt, several models, and pick the cut that holds attention.
Image to motion
Animate a still, a moodboard or a storyboard frame into a few seconds of motion for a pitch.
How it works
- 1
Pick a model
Every route on this page has a model ID, an accepted input mode and a published rate.
- 2
Send a request
POST the prompt, and any frame or reference assets, to the videos endpoint with your Felo key.
- 3
Poll and download
The endpoint returns a task ID; poll it and download the finished file when it is ready.
curl https://openapi.felo.ai/api/v1/videos/generations \
-H "Authorization: Bearer $FELO_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"mischief-1","prompt":"A slow push-in on a desk at sunrise"}'
How billing works
Video is priced per second of generated output on most routes, or per output token where the provider prices that way. The table above shows each model's live rate.
- Rates come from the default channel and are deducted from prepaid USD credits.
- Top up from $1; there is no subscription and no monthly minimum.
- A trial price can change, so check the model page when you send the request.
Try it online, no code
Generate a video in the browser
No code and no key: describe the shot in Felo's AI Video Generator and download the result.
Open AI Video GeneratorTest any model in the Playground
Pick a model, send a request and read the response in the browser. No local setup, no install.
Open PlaygroundModel cases
Browse all model casesReal outputs and prompts from models in this lineup.
Why teams use Felo
Felo carries the video models people want in 2027 — Seedance, Kling, Veo, Wan, MiniMax and Grok Imagine — under one key and one bill, so you compare the full lineup instead of onboarding provider by provider.
- Every model in the 2027 lineup is callable with the same key and the same request shape.
- Video, image, chat and data APIs share the same key and the same balance.
- Prices, specs and availability are read from the live catalog, not from a stale list.
Frequently asked questions
What is the best AI video model API?
There is no single best model. Felo lists text-to-video, image-to-video and reference-driven routes side by side with live per-second prices, resolutions and clip limits, so the right choice depends on the shot, the resolution and the budget.
Can I use Seedance, Kling or Veo without vendor accounts?
Yes. Seedance, Kling, Veo, Wan, MiniMax and Grok Imagine routes on this page are called through a Felo key. You top up service credits and pay per call instead of opening an account with each provider.
How much does AI video generation cost per second?
It varies by model and resolution. Routes in the catalog are priced per second of output, per output token, or per request; the table shows the current rate for each model, and the model page is the source of truth when you send a request.
Can I generate an AI video without code?
Yes. Use Felo's AI Video Generator in the browser to describe a shot and download the video, or open the Playground to call a specific model with your own prompt and settings.
What input does a video model accept?
Most routes accept a text prompt; many also accept a first and last frame, or reference images and video. The input-mode tags under each model show what it takes, and text-only routes say so.
Is there a free trial for the video API?
Trial availability changes by model and is shown in the catalog; there is no recurring free allowance. Create a key and top up from $1 to test routes beyond a trial, and check the model page for the current rate.
What are the best AI video models in 2027?
Felo lists the full 2027 lineup — 14 video models from 8 providers, including Seedance, Kling, Veo, Wan, MiniMax and Grok Imagine — with live rates; the best pick depends on the shot and the resolution you need.
Call your first video model
Create a key, pick a model ID and send one request. The Playground is there if you would rather click than code.