4 AI models · 15 ranked · updated this week
Best AI Video Generators, according to AI (2026).
There is no single answer. The models crown 2 different leaders this week. Runway holds the consensus #1 at a score of 50, but the model-by-model record below shows a genuinely contested category.
This guide is built from the recorded answers of ChatGPT, Claude, Gemini, Perplexity to the real questions buyers ask about ai video generators. We logged 1,690 of them this period, covering 15 tools. We report what the models said, in the order they said it. Nobody paid to be here, and we don't add opinions of our own.
Prefer the raw board? See the full model-by-model ranking. Every rank, every score, every week.
The ranked list
Runway holds the consensus #1 at a score of 50, with ChatGPT and Claude and Gemini placing it first outright.
The models disagree about it more than most: #1 on ChatGPT but #11 on Perplexity, a spread of 10.
It entered the ranking this week. A new name in the models' answers.
It is the most contested name in this category. It carries the widest cross-model disagreement on the board.
Claude and Perplexity place Google Veo at #1, even though it sits at #2 on consensus. That is a genuine split in the record.
It entered the ranking this week. A new name in the models' answers.
Claude places Kling AI at #1, even though it sits at #3 on consensus. That is a genuine split in the record.
The models disagree about it more than most: #1 on Claude but #4 on ChatGPT, a spread of 3.
It entered the ranking this week. A new name in the models' answers.
Claude places Runway Gen-4.5 at #1, even though it sits at #4 on consensus. That is a genuine split in the record.
The models disagree about it more than most: #1 on Claude but #8 on ChatGPT, a spread of 7.
It entered the ranking this week. A new name in the models' answers.
Claude places Luma Dream Machine at #1, even though it sits at #5 on consensus. That is a genuine split in the record.
The models disagree about it more than most: #1 on Claude but #11 on Perplexity, a spread of 10.
It entered the ranking this week. A new name in the models' answers.
Claude places OpenAI Sora at #1, even though it sits at #6 on consensus. That is a genuine split in the record.
The models disagree about it more than most: #1 on Claude but #8 on ChatGPT, a spread of 7.
It entered the ranking this week. A new name in the models' answers.
Claude places Adobe Firefly Video at #1, even though it sits at #7 on consensus. That is a genuine split in the record.
The models disagree about it more than most: #1 on Claude but #11 on Perplexity, a spread of 10.
It entered the ranking this week. A new name in the models' answers.
Claude places Luma Dream Machine / Ray at #1, even though it sits at #8 on consensus. That is a genuine split in the record.
The models disagree about it more than most: #1 on Claude but #8 on ChatGPT, a spread of 7.
It entered the ranking this week. A new name in the models' answers.
Claude places Seedance 2.0 at #1, even though it sits at #9 on consensus. That is a genuine split in the record.
The models disagree about it more than most: #1 on Claude but #11 on Perplexity, a spread of 10.
It entered the ranking this week. A new name in the models' answers.
Claude places Synthesia at #1, even though it sits at #10 on consensus. That is a genuine split in the record.
The models disagree about it more than most: #1 on Claude but #8 on ChatGPT, a spread of 7.
It entered the ranking this week. A new name in the models' answers.
Claude places Hailuo AI at #1, even though it sits at #11 on consensus. That is a genuine split in the record.
The models disagree about it more than most: #1 on Claude but #11 on Perplexity, a spread of 10.
It entered the ranking this week. A new name in the models' answers.
Claude places Seedance at #1, even though it sits at #12 on consensus. That is a genuine split in the record.
The models disagree about it more than most: #1 on Claude but #8 on ChatGPT, a spread of 7.
It entered the ranking this week. A new name in the models' answers.
Claude places HeyGen at #1, even though it sits at #13 on consensus. That is a genuine split in the record.
The models disagree about it more than most: #1 on Claude but #8 on Perplexity, a spread of 7.
It entered the ranking this week. A new name in the models' answers.
Claude places Pika at #1, even though it sits at #14 on consensus. That is a genuine split in the record.
The models disagree about it more than most: #1 on Claude but #9 on Perplexity, a spread of 8.
It entered the ranking this week. A new name in the models' answers.
Claude places Hailuo Minimax / Minimax 2.3 at #1, even though it sits at #15 on consensus. That is a genuine split in the record.
The models disagree about it more than most: #1 on Claude but #10 on Perplexity, a spread of 9.
It entered the ranking this week. A new name in the models' answers.
Questions people ask.
What is the best ai video generators according to AI?
Runway holds the consensus #1 this week with a score of 50 out of 100, ahead of Google Veo. The models don't fully agree: 2 different brands are crowned #1 across the four models.
Which ai video generator does ChatGPT recommend first?
ChatGPT currently places Runway at #1 for ai video generators, per this week's recorded answers.
Which ai video generator does Claude recommend first?
Claude currently places Runway at #1 for ai video generators, per this week's recorded answers.
How are these rankings measured?
We ask each model the same buying questions every week (1,690 queries this period), record the full answers, and score each named brand 0-100 by how early and how consistently it appears. Every question runs through the official model APIs, with web search on, not through the consumer chat apps, so nothing is personalized to a user. Each model is scored independently; the consensus blends all 4.
Do the AI models agree with each other?
Not fully. The four models crown 2 different #1s this week, and several brands carry wide spreads. That is the gap between their best and worst model rank. That disagreement is a finding, not noise: different models read different sources.
This is the guide. The record has more.
The full ranking shows every model’s column side by side, 12 weeks of movement, and the methodology behind every number.