The creative model reference

AI video models for film

Compare reference rankings for cinematic storytelling and film production tasks.

September 2026 benchmark snapshot · Curated by Oakgen. Data source and limits.

video / Film

The scores, in perspective.

11 models · 95% intervals

Swipe to compare all models →

Oakgen Arena · v1.0

11,132

Seedance 2.5

21,080

Seedance 2.0

31,039

Wan 3.0

41,038

Seedance 2.0 Mini

51,015

Google Omni Flash

61,000

MiniMax H3Anchor

7991

Flux 3 Video

8977

HappyHorse 1.1

9946

Kling 3.0 Omni

10934

Grok Imagine 1.5

11909

PixVerse V6

↥ 95% confidence interval · Anchor interval not reported.11 ranked models

Relative scores on a zero-based display scale. Compare within this board only. Values are rounded for display.

September 2026 benchmark snapshot · Curated by Oakgen. Data source and limits.

Published reference / v1.0

Film leaderboard

Film video rankings · Benchmark index 1.0
Rank	Model	Overall score	95% confidence interval	Vote appearances	Distinct judges
01	Seedance 2.5ByteDance Criterion scores Acting/performance1,148.495% CI 1,085.5 to 1,215.8 Audio quality1,178.195% CI 1,128.4 to 1,228.7Reused fit from master-video Camera & lighting control1,222.295% CI 1,150.7 to 1,314.5 Consistency1,194.795% CI 1,151.8 to 1,243.3Reused fit from master-video Multi-shot continuity1,179.695% CI 1,107.6 to 1,270.1 Physics & motion1,103.695% CI 1,068.6 to 1,137.4Reused fit from master-video Prompt adherence1,052.695% CI 1,014.6 to 1,088.4Reused fit from master-video Realism1,149.695% CI 1,065.7 to 1,246.1 Style understanding977.195% CI 886.6 to 1,072.8 Subjective aesthetics1,096.495% CI 1,061.3 to 1,131.1Reused fit from master-video	1,132.3LEAD	1,108.6 to 1,161.8	4,896	Not reported
02	Seedance 2.0ByteDance Criterion scores Acting/performance1,152.895% CI 1,083.7 to 1,229.6 Audio quality1,102.895% CI 1,063.5 to 1,140.8Reused fit from master-video Camera & lighting control1,160.595% CI 1,102.5 to 1,228 Consistency1,060.895% CI 1,022.7 to 1,099Reused fit from master-video Multi-shot continuity1,138.295% CI 1,086.1 to 1,195.9 Physics & motion1,036.195% CI 1,002.3 to 1,069.3Reused fit from master-video Prompt adherence1,050.595% CI 1,017.4 to 1,084.1Reused fit from master-video Realism1,011.395% CI 937.9 to 1,087.2 Style understanding1,013.695% CI 935.7 to 1,095.4 Subjective aesthetics971.895% CI 935.7 to 1,010.3Reused fit from master-video	1,080#2	1,058.2 to 1,104.1	4,930	Not reported
03	Wan 3.0Alibaba Criterion scores Acting/performance948.895% CI 872.8 to 1,022.8 Audio quality1,059.695% CI 1,022.9 to 1,100.1Reused fit from master-video Camera & lighting control1,135.495% CI 1,063.4 to 1,222.2 Consistency1,121.595% CI 1,080.6 to 1,165.8Reused fit from master-video Multi-shot continuity1,066.895% CI 1,010.8 to 1,134.1 Physics & motion1,026.795% CI 993 to 1,058.4Reused fit from master-video Prompt adherence980.895% CI 940.2 to 1,021.2Reused fit from master-video Realism1,028.995% CI 958.8 to 1,113 Style understanding998.495% CI 913.3 to 1,090 Subjective aesthetics1,044.395% CI 1,006.5 to 1,083.1Reused fit from master-video	1,038.9#3	1,016 to 1,066.8	4,903	Not reported
04	Seedance 2.0 MiniByteDance Criterion scores Acting/performance1,076.195% CI 1,016.3 to 1,130.5 Audio quality1,127.895% CI 1,085.5 to 1,172.8Reused fit from master-video Camera & lighting control1,11495% CI 1,028.1 to 1,227.1 Consistency1,03495% CI 997.3 to 1,069.8Reused fit from master-video Multi-shot continuity1,030.995% CI 969.2 to 1,089.1 Physics & motion1,024.495% CI 991.6 to 1,059.6Reused fit from master-video Prompt adherence971.495% CI 937.1 to 1,002.7Reused fit from master-video Realism1,026.595% CI 949.6 to 1,115.3 Style understanding953.195% CI 894.4 to 1,019.6 Subjective aesthetics1,008.595% CI 972.9 to 1,048.3Reused fit from master-video	1,038.1	1,016.9 to 1,063.2	4,958	Not reported
05	Google Omni FlashGoogle Criterion scores Acting/performance991.895% CI 927.3 to 1,058.4 Audio quality1,108.895% CI 1,065.6 to 1,152.5Reused fit from master-video Camera & lighting control1,127.995% CI 1,065.5 to 1,197.2 Consistency995.495% CI 956 to 1,033.9Reused fit from master-video Multi-shot continuity969.295% CI 910.8 to 1,022.9 Physics & motion98395% CI 949.1 to 1,020.3Reused fit from master-video Prompt adherence977.295% CI 942.3 to 1,010.9Reused fit from master-video Realism1,004.495% CI 931.7 to 1,082.8 Style understanding950.895% CI 884.5 to 1,025.4 Subjective aesthetics1,081.695% CI 1,041.6 to 1,123.6Reused fit from master-video	1,014.9	994.2 to 1,038.4	4,904	Not reported
06	MiniMax H3MiniMax Criterion scores Acting/performance1,00095% CI Anchor · interval not reported Audio quality1,00095% CI Anchor · interval not reportedReused fit from master-video Camera & lighting control1,00095% CI Anchor · interval not reported Consistency1,00095% CI Anchor · interval not reportedReused fit from master-video Multi-shot continuity1,00095% CI Anchor · interval not reported Physics & motion1,00095% CI Anchor · interval not reportedReused fit from master-video Prompt adherence1,00095% CI Anchor · interval not reportedReused fit from master-video Realism1,00095% CI Anchor · interval not reported Style understanding1,00095% CI Anchor · interval not reported Subjective aesthetics1,00095% CI Anchor · interval not reportedReused fit from master-video	1,000Anchor	Anchor · interval not reported	4,938	Not reported
07	Flux 3 VideoBlack Forest Labs Criterion scores Acting/performance1,001.395% CI 937.9 to 1,056.9 Audio quality1,078.195% CI 1,036.8 to 1,122.5Reused fit from master-video Camera & lighting control1,048.295% CI 964 to 1,141.2 Consistency96395% CI 929.3 to 994.1Reused fit from master-video Multi-shot continuity99795% CI 924.4 to 1,060.9 Physics & motion994.695% CI 963.3 to 1,028.8Reused fit from master-video Prompt adherence979.195% CI 947.2 to 1,011.5Reused fit from master-video Realism93795% CI 882.5 to 988.9 Style understanding942.795% CI 861.9 to 1,016.4 Subjective aesthetics1,000.995% CI 963.9 to 1,038.2Reused fit from master-video	990.6	968.9 to 1,014.1	4,988	Not reported
08	HappyHorse 1.1HappyHorse Criterion scores Acting/performance973.795% CI 899.1 to 1,043 Audio quality1,000.195% CI 962.1 to 1,042.4Reused fit from master-video Camera & lighting control1,066.295% CI 992.2 to 1,151.3 Consistency994.495% CI 956.8 to 1,030.8Reused fit from master-video Multi-shot continuity994.395% CI 931.9 to 1,057 Physics & motion960.295% CI 928.1 to 993.1Reused fit from master-video Prompt adherence944.895% CI 903.6 to 982.1Reused fit from master-video Realism911.595% CI 821.6 to 994.5 Style understanding939.195% CI 862.8 to 1,019.1 Subjective aesthetics985.695% CI 950 to 1,022.2Reused fit from master-video	977	953.4 to 1,001.1	4,906	Not reported
09	Kling 3.0 OmniKuaishou Criterion scores Acting/performance938.995% CI 874.3 to 1,003.5 Audio quality1,053.595% CI 1,007.9 to 1,099.6Reused fit from master-video Camera & lighting control1,027.495% CI 957.5 to 1,105.1 Consistency1,041.995% CI 999 to 1,083.2Reused fit from master-video Multi-shot continuity92795% CI 861.4 to 986.6 Physics & motion884.495% CI 842.6 to 921.2Reused fit from master-video Prompt adherence962.995% CI 920.5 to 1,002.9Reused fit from master-video Realism915.495% CI 855.9 to 977.6 Style understanding851.495% CI 777 to 934.7 Subjective aesthetics954.795% CI 917.5 to 990.5Reused fit from master-video	946.3	923.7 to 968.4	4,924	Not reported
10	Grok Imagine 1.5xAI Criterion scores Acting/performance909.695% CI 828.7 to 988.1 Audio quality904.795% CI 863.9 to 944.8Reused fit from master-video Camera & lighting control1,03695% CI 947.3 to 1,132.2 Consistency1,050.895% CI 1,014 to 1,090.3Reused fit from master-video Multi-shot continuity84495% CI 761.1 to 916.2 Physics & motion925.195% CI 885.9 to 958.7Reused fit from master-video Prompt adherence94395% CI 906.9 to 982.2Reused fit from master-video Realism949.695% CI 875.4 to 1,021.8 Style understanding868.595% CI 792.1 to 950.2 Subjective aesthetics99395% CI 963.3 to 1,027.5Reused fit from master-video	934.1	907.1 to 959.9	4,894	Not reported
11	PixVerse V6PixVerse Criterion scores Acting/performance816.395% CI 750 to 888.4 Audio quality977.295% CI 940.8 to 1,012.5Reused fit from master-video Camera & lighting control1,027.295% CI 930.3 to 1,131.1 Consistency990.195% CI 946.9 to 1,031.7Reused fit from master-video Multi-shot continuity846.895% CI 776.8 to 915.2 Physics & motion935.395% CI 893.1 to 972.7Reused fit from master-video Prompt adherence88495% CI 838.3 to 926Reused fit from master-video Realism894.295% CI 822.3 to 966.8 Style understanding866.895% CI 786.1 to 945.7 Subjective aesthetics979.995% CI 943.6 to 1,014.5Reused fit from master-video	909.2	883.7 to 933.9	4,931	Not reported

Ranks are the source’s published positions. Vote appearances are not unique people and must not be added across boards. Scores are rounded for display.

September 2026 benchmark snapshot · Curated by Oakgen. Data source and limits.

From comparison to creation

Choose for your brief.

Test a sequence of related shots, not only a striking single clip. Check character continuity, lighting and the camera language you need. Review the output at the intended edit length before building a full sequence around a model.

Explore Oakgen's model directory

A source listing does not confirm current availability in Oakgen.

How this board is composed

Criteria and weights

Acting/performance14%
Multi-shot continuity14%
Physics & motion6%
Camera & lighting control14%
Style understanding14%
Realism14%
Consistency6%
Audio quality6%
Prompt adherence6%
Subjective aesthetics6%

Read the methodology