Best Open Source AI Video Generators (2026)

Summary: 8 open-source AI video models compared as of April 29, 2026. HappyHorse 1.0 is API-available, but not a verified self-hostable release yet. Best API value: LTX-2 Pro at $0.06/sec with 1080p + audio + lip-sync (LTX-2 Community License). Cheapest API: HunyuanVideo 1.5 at $0.02/sec on WaveSpeed. Best for consumer GPUs: FramePack runs on 6GB VRAM (RTX 3060, ~$300). Self-hosting breaks even at ~200 clips/month for some open-weight workflows. The quality gap has narrowed, but not every announced open-weight model is production-usable yet.

Open-source AI video has become strong enough for real production workflows, but the most important detail is whether a model is actually usable today. HappyHorse 1.0 is a high-interest API-available model, but we cannot verify public weights or self-hosting rights yet; see the HappyHorse availability tracker. Of the 28 model records we track, several are verified open-weight options, and the cheapest costs just $0.02/sec via API.

We compared every open-source AI video model on license terms, hardware requirements, API pricing, and self-hosting viability. Whether you want the cheapest API option or full control by running models on your own GPUs, this guide covers the complete landscape.

Prices verified: April 29, 2026.

All Open-Source AI Video Models

Model Deep Dives

HappyHorse 1.0 — API-Available, Not Open Source

HappyHorse 1.0 is a high-interest model because public reporting attributes strong benchmark performance and multilingual lip-sync to it. But availability matters more than launch claims: FAL and Replicate now verify provider API access and generation pricing, but we cannot currently verify downloadable weights, inference code, license terms, or a mature public repo.

The practical guidance is simple: test HappyHorse through provider APIs if you want to compare quality, but do not count it as a production self-hosting option until public weights, license terms, and runnable inference are verified. We track those checks in the HappyHorse availability page.

Wan 2.7 — Hosted API and Open-Weight Watchlist

Wan 2.7 from Alibaba is a hosted/API model in this catalog, not a verified self-hostable release. At 27B parameters using a Mixture-of-Experts (MoE) architecture, it supports four generation modes: text-to-video, image-to-video, video-to-video, and audio-driven generation. API pricing starts at $0.10/sec on FAL.ai. Until public weights, inference code, and license terms are verified, keep Wan 2.7 in the open-weight watchlist bucket rather than the self-hostable bucket.

LTX-2 Pro — Best Balance of Quality, Price, and Features

LTX-2 Pro is the open-source model we recommend most often. At $0.06/secwith audio included, it’s the cheapest 1080p+audio option on the market. It uses the LTX-2 Community License, has 14B parameters, supports native 4K ($0.24/sec at 4K on the verified FAL endpoint), and includes lip-sync capability. Its 50fps output is smooth enough for slow-motion content.

HunyuanVideo 1.5 — Cheapest API Option

HunyuanVideo 1.5 from Tencent is the absolute cheapest AI video model via API at $0.02/sec on WaveSpeed. At 8.3B parameters and 14GB VRAM, it’s also one of the most accessible for self-hosting. The trade-off is that current provider docs keep the default generation range around 480p-720p, with 1080p treated as super-resolution/upscale, and no native audio. Best for high-volume generation where quality is secondary to cost.

FramePack — Best for Consumer Hardware

FramePack is the standout for self-hosting on consumer GPUs. It runs on just 6GB VRAM (an RTX 3060 at ~$300) and its local project documents 60-second / 1,800-frame generation at 30fps. The hosted FAL endpoint is narrower: image-to-video, 480p/720p tiers, and a 180-frame cap per generation. There is no native audio. The API price on FAL.ai starts at $0.033/sec, but self-hosting on a consumer GPU eliminates per-second costs entirely.

Mochi 1 — Apache 2.0 Pioneer

Mochi 1 from Genmo was one of the first high-quality open-source video models. At 10B parameters with an Apache 2.0 license, it produces 480p output. API pricing is $0.40 per clip (flat rate). Mochi 1 laid the groundwork for the open-source video generation ecosystem, though newer models like LTX-2 Pro and Wan 2.7 have surpassed it on quality.

CogVideoX-5B — Lightweight Option

CogVideoX-5B is the smallest model in this comparison at 5B parameters. Its custom license permits commercial use. At 16GB VRAM, it runs on an RTX 4090 or equivalent. Output is 480p with API pricing at $0.20 per clip. Best suited as a starting point for teams exploring self-hosted video generation with limited GPU resources.

SkyReels V4 — Hosted/API Until V4 Release Is Verified

SkyReels V4 from Skywork AI is a strong multimodal video candidate with reported API access at $0.12/sec with audio included, but we do not verify public V4 weights, runnable inference code, or license terms yet. Treat it as hosted/API coverage, not a self-hosting recommendation, until those release artifacts are public.

Self-Host Economics: When to Run Your Own GPU

Self-hosting eliminates per-second API costs but introduces fixed infrastructure costs. Here’s when it makes financial sense.

Breakeven estimates assume 5-second clips with generation overhead. Actual numbers depend on generation speed and GPU utilization.

Decision Framework

• Under 200 clips/month: Use API pricing. Zero infrastructure overhead means the per-second cost is absorbed by operational simplicity.
• 200–500 clips/month:Self-hosting on a consumer GPU (RTX 3060 for FramePack, RTX 4090 for HunyuanVideo) breaks even. One-time hardware cost pays for itself within 1–3 months.
• 500+ clips/month:Dedicated pro GPU (A5000/A6000) or cloud H100 rental becomes strongly cost-effective. Running LTX-2 Pro on an A6000 saves $0.06/sec on every clip — at 500 clips that’s $150/month in API savings.
• Enterprise scale: Cloud H100 rental at ~$2/hr supports large verified open-weight models. At continuous utilization, this can be cheaper than API pricing for high-quality models, but it only applies to models with public weights and runnable inference code.

Open Source vs Commercial: Quality Comparison

Key Insight: The Quality Gap Has Closed

The narrative that commercial AI video models are fundamentally better than open-source alternatives is no longer true. Verified open-weight models now cover real workflows, and FramePack proves you can generate AI video on a $300 consumer GPU. But API-available non-open models like HappyHorse 1.0 and hosted watchlist models like Wan 2.7 should stay out of the open-source bucket until their public weights and license terms are verified; start with the HappyHorse availability tracker before treating it as an open-source option.

What commercial models still offer is convenience and features: Kling v3’s multi-shot generation, Veo 3.1’s premium lip-sync, and Runway Gen-4.5’s cinematic aesthetics are not yet matched by open-source alternatives. But for raw quality-per-dollar, open source leads.

For complete API pricing across all models, see our AI Video Pricing Guide, or compare providers with our API Provider Guide.

FAQ