Skip to content

Rankings

Most used AI models this week

The AI models people on FunAI ran most this week, for video, image, audio, 3D and text. Ranked by share of successful runs, with each model’s price and how it moved since last week.

Week of 12–19 November 2026

Video models

The top 2 by share of this week’s video runs.

Most used video models this week
#Change since last weekModelShare
1NewNew in the ranking this weekSeedance-2.0-fastByteDance · about $0.254 / second25%+ of runs
2NewNew in the ranking this weekSeedance 2.5ByteDance · about $0.232 / second10–25% of runs

Image models

The top 5 by share of this week’s image runs.

Most used image models this week
#Change since last weekModelShare
11Up 1 place since last weekNano Banana 2Gemini · $0.084 / image25%+ of runs
2NewNew in the ranking this weekSDXL LightningByteDance · $0.00105 / image10–25% of runs
3NewNew in the ranking this weekZ Image TurboAlibaba · $0.00893 / image5–10% of runs
4NewNew in the ranking this weekFlux KreaBlack Forest Labs · $0.00893 / image5–10% of runs
54Down 4 places since last weekQwen ImageAlibaba · $0.00105 / image5–10% of runs

Audio models

The top 10 by share of this week’s audio runs.

Not enough audio runs this week to rank models yet. Browse audio models

3D models

The top 10 by share of this week’s 3D runs.

Not enough 3D runs this week to rank models yet. Browse 3D models

Text models

The top 2 by share of this week’s text runs.

Most used text models this week
#Change since last weekModelShare
1NewNew in the ranking this weekGemini 3.1 Flash LiteGemini · about $0.263 in / $1.57 out per 1M tokens25%+ of runs
2NewNew in the ranking this weekGPT-5 NanoOpenAI · about $0.052 in / $0.42 out per 1M tokens25%+ of runs

How these figures are made

Ranking
Ranked by successful customer runs in the last 7 days, per output type. Tool runs, staff and test runs, and helper steps are not counted. Variants of a model stored under older names are merged into the catalog model. Share is the model's part of all successful runs of that output type, shown as a band.
Success rate and time to result
Last 30 days of customer runs. Success rate = successful runs / (successful runs + runs that failed on our side or the model's side). Runs refused by a safety filter, runs with invalid settings or missing input, runs stopped by the customer and runs without enough credits are not counted as failures. Time to result = from submitting to the finished result, including any queue wait; we show the median. Figures are hidden until a model has 20 runs in the window.
Share
Each model’s share of the week’s successful runs of its type, as a band. We publish ranks and bands, never the number of runs.