SlopTV open data

The dataset behind SlopTV, an independent benchmark and price tracker for AI video generation. Every model runs the same 40 prompts, every clip is scored, and prices are read twice a day across 17 platforms and APIs.

Current leaderboard

#	Model	Maker	Overall	95% CI
1	Kling 3.0 Turbo	Kuaishou	4.84	±0.11
2	FLUX 3	Black Forest Labs	4.84	±0.10
3	Seedance 2.0	ByteDance	4.84	±0.15
4	Google Gemini Omni	Google	4.82	±0.12
5	MiniMax H3	MiniMax	4.79	±0.22
6	Seedance 2.5	ByteDance	4.79	±0.15
7	Vivix W1	Vivix	4.77	±0.15
8	OpenAI Sora 2	OpenAI	4.76	±0.17
9	Kling 4.0	Kuaishou	4.76	±0.10
10	Wan 3.0	Alibaba	4.75	±0.12
11	Grok Video	xAI	4.72	±0.16
12	Google Veo 3.1	Google	4.63	±0.26
13	Runway Gen-4.5	Runway	4.57	±0.16
14	PixVerse v5.5	PixVerse	4.52	±0.18
15	Kling 3.0	Kuaishou	4.47	±0.20
16	MiniMax Hailuo 02	MiniMax	3.90	±0.47

Live version with every clip: sloptv.co/leaderboard. Where each model is cheapest: the price radar.

Download

leaderboard.csv (16 rows): One row per model: overall score, 95% confidence interval, sub-scores and the score in each test category.
scores.csv (698 rows): One row per scored clip: model, prompt, category, round and every judge score, with a link to the clip.
prices.csv (13,081 rows): Current published prices on every platform and API we track, normalised to USD per clip and per second.

Column definitions are in the repository README. Methodology: sloptv.co/methodology.

License

CC BY 4.0. Free to use, including commercially, with attribution and a link: Source: <a href="https://sloptv.co">SlopTV</a>