GoogleVideo model
Veo 3.1 Fast
In text to video, Veo 3.1 Fast is in place 11 of 23, with image to video in place 11. A 5 s clip costs about $0.50.
1,385Arena ratinghigher is better
1,362Arena ratinghigher is better
$0.50$0.10 per secondlower is better
8slongest single cliphigher is better
31out of 100-4 vs. last month
Summary
At $0.10 per second of video it is fairly cheap; a typical video model costs $0.12.
Veo 3.1 Fast is the faster variant beside Veo 3.1. Both variants share native audio, reference images, scene extension, and first-to-last-frame transitions, reaching 1080p and 8 seconds.
Google also schedules the veo-3.1-fast-generate-preview endpoint for shutdown on October 22, 2026, and recommends Gemini Omni 1.1 Flash as the replacement. Fast is therefore an ending preview access path.
Veo 3.1 Fast benchmarks
Image to video arena
How people rate generated videos in blind comparisons, Elo points · higher is better
20 of 21 models
as of September 22, 2026 · Source: LMArena
Text to video arena
How people rate generated videos in blind comparisons, Elo points · higher is better
20 of 23 models
as of September 22, 2026 · Source: LMArena
Show all 104 individual results from 19 sources
Page 1 of 9
Published benchmarks
LMArena Text to Video
Elo ratings summarize pairwise comparisons within this leaderboard. They are not percentages. The scale does not express absolute quality or linear gaps between models.
higher is betterAs of November 5, 2026Scale: 0 Elo - 1,516 Elo
Methodology and context
- Metric
- elo-rating
- Scale
- No upper bound is documented. The chart spans the recorded values and zero.
- Participants
- 23
- Sample
- 39,484
LMArena Image to Video
Elo ratings summarize pairwise comparisons within this leaderboard. They are not percentages. The scale does not express absolute quality or linear gaps between models.
higher is betterAs of November 5, 2026Scale: 0 Elo - 1,495 Elo
Methodology and context
- Metric
- elo-rating
- Scale
- No upper bound is documented. The chart spans the recorded values and zero.
- Participants
- 21
- Sample
- 99,804
Measurements without matching peer values
These measurements have no matching peer values under the same test conditions. Their original values and sources remain available here.
VideoASMR-Bench Fakeness_G, ImgText2Vid, ASMR image suite used (Show measurement, test conditions, and source)
- Source value
- 12.54
- Score
- 12.54
- Metric
- Fakeness_G, ASMR fake-video detection rate averaged over evaluation VLMs; lower is more realistic
- Unit
- %
- Benchmark version
- VideoASMR-Bench arXiv:2512.13281v4, Table 3
- Direction
- Lower is better
- Source type
- independent-evaluator
- Evaluator
- VideoASMR-Bench paper, arXiv:2512.13281v4
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. VideoASMR-Bench Table 3, generation model Veo-3.1-fast, ImgText2Vid, ASMR image suite used; the table evaluates the source's ASMR prompt/image suite and its fake-video detection setting. The paper does not state an exact immutable API route or serving revision for this generation row.
- Context
- Fakeness_G is the proportion of generated fake clips judged non-real, averaged over the evaluated video-understanding models; the source explicitly says lower means more realistic. This is an ASMR-focused perception task, not a general video-quality score. The paper reports neither a per-row sample count nor a confidence interval.
VideoASMR-Bench Fakeness_G, ImgText2Vid with audio output, ASMR image suite used (Show measurement, test conditions, and source)
- Source value
- 15.05
- Score
- 15.05
- Metric
- Fakeness_G, ASMR fake-video detection rate averaged over evaluation VLMs; lower is more realistic
- Unit
- %
- Benchmark version
- VideoASMR-Bench arXiv:2512.13281v4, Table 3
- Direction
- Lower is better
- Source type
- independent-evaluator
- Evaluator
- VideoASMR-Bench paper, arXiv:2512.13281v4
- Status
- active
- Retrieved at
- 2026-10-17
- Methodology
- Independent evaluation. VideoASMR-Bench Table 3, generation model Veo-3.1-fast, ImgText2Vid with audio output, ASMR image suite used; the table evaluates the source's ASMR prompt/image suite and its fake-video detection setting. The paper does not state an exact immutable API route or serving revision for this generation row.
- Context
- Fakeness_G is the proportion of generated fake clips judged non-real, averaged over the evaluated video-understanding models; the source explicitly says lower means more realistic. This is an ASMR-focused perception task, not a general video-quality score. The paper reports neither a per-row sample count nor a confidence interval.
AA Video Arena Elo (Show measurement, test conditions, and source)
- Source value
- 965
- Score
- 965
- Metric
- pairwise human-preference Elo
- Unit
- elo
- Benchmark version
- AA-Video-T2V v2.1
- Category
- text-to-video
- Direction
- Higher is better
- Sample
- 3985
- Confidence interval
- 17
- Lower confidence bound
- 948
- Upper confidence bound
- 982
- Source type
- independent-evaluator
- Evaluator
- Artificial Analysis AA-Video-T2V v2.1
- Status
- superseded
- Retrieved at
- 2026-10-16
- Methodology
- Independent evaluation. Text-to-video with audio; exact source row Veo 3.1 Fast.
- Context
- The published comparison includes audio; it is not a silent-video-only score.
AA Video Arena Elo (Show measurement, test conditions, and source)
- Source value
- 1017
- Score
- 1017
- Metric
- pairwise human-preference Elo
- Unit
- elo
- Benchmark version
- AA-Video-I2V v1.1
- Category
- image-to-video
- Direction
- Higher is better
- Sample
- 4826
- Confidence interval
- 16
- Lower confidence bound
- 1001
- Upper confidence bound
- 1033
- Source type
- independent-evaluator
- Evaluator
- Artificial Analysis AA-Video-I2V v1.1
- Status
- active
- Retrieved at
- 2026-10-16
- Methodology
- Independent evaluation. Image-to-video with audio; exact source row Veo 3.1 Fast.
- Context
- The published comparison includes audio; it is not a silent-video-only score.
Physics-IQ Verified leaderboard score (Show measurement, test conditions, and source)
- Source value
- 29.96
- Score
- 29.96
- Metric
- Physics-IQ Verified physical-understanding score on a 0-100 point scale
- Unit
- points
- Benchmark version
- Physics-IQ Verified leaderboard
- Category
- image-to-video
- Direction
- Higher is better
- Source type
- official-leaderboard
- Evaluator
- Physics-IQ Verified GitHub leaderboard
- Status
- active
- Retrieved at
- 2026-10-16
- Methodology
- Official benchmark leaderboard. Image-to-video (I2V); unmarked entry uses the model-specific best-practice prompt template. The leaderboard row does not disclose output resolution, duration, or a per-model sample count.
- Context
- This measures physical understanding on a 0-100 point scale, not general video quality. The benchmark paper does not evaluate this newer model, so no generation controls are imported from it. The leaderboard displays ± notation; no confidence interval or row-specific run count is inferred.
Video Models AI-judge benchmark (Show measurement, test conditions, and source)
- Source value
- 3.69
- Score
- 3.69
- Metric
- Weighted overall AI-judge quality score (0-5)
- Unit
- points
- Benchmark version
- Video Models leaderboard updated 2026-10-15
- Category
- video-generation
- Direction
- Higher is better
- Sample
- 64
- Source type
- vendor-reported
- Evaluator
- Video Models by Wiro AI
- Status
- active
- Retrieved at
- 2026-10-16
- Methodology
- Vendor-reported result. Exact Wiro API route google/veo3-1-fast. Published method states one API run per prompt using model defaults, first successful output, a 5-second duration target, and the resolution closest to 720p. This route documentation lists 4-8s; 720p or 1080p; maximum 8s. The CSV does not expose the exact per-prompt overrides or criterion-specific denominators.
- Context
- Wiro AI is a commercial model gateway. Scores use its weighted 0-5 AI-judge rubric, not a blind human pairwise preference panel. No per-criterion N or confidence interval is published. This is task-specific AI-judge evidence, not broad independent quality proof.
Video Models AI-judge benchmark (Show measurement, test conditions, and source)
- Source value
- 3.43
- Score
- 3.43
- Metric
- Prompt adherence AI-judge criterion score (0-5); rubric weight 21%
- Unit
- points
- Benchmark version
- Video Models leaderboard updated 2026-10-15
- Category
- video-generation
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Video Models by Wiro AI
- Status
- active
- Retrieved at
- 2026-10-16
- Methodology
- Vendor-reported result. Exact Wiro API route google/veo3-1-fast. Published method states one API run per prompt using model defaults, first successful output, a 5-second duration target, and the resolution closest to 720p. This route documentation lists 4-8s; 720p or 1080p; maximum 8s. The CSV does not expose the exact per-prompt overrides or criterion-specific denominators.
- Context
- Wiro AI is a commercial model gateway. Scores use its weighted 0-5 AI-judge rubric, not a blind human pairwise preference panel. No per-criterion N or confidence interval is published. This is task-specific AI-judge evidence, not broad independent quality proof.
Video Models AI-judge benchmark (Show measurement, test conditions, and source)
- Source value
- 3.52
- Score
- 3.52
- Metric
- Motion and consistency AI-judge criterion score (0-5); rubric weight 19%
- Unit
- points
- Benchmark version
- Video Models leaderboard updated 2026-10-15
- Category
- video-generation
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Video Models by Wiro AI
- Status
- active
- Retrieved at
- 2026-10-16
- Methodology
- Vendor-reported result. Exact Wiro API route google/veo3-1-fast. Published method states one API run per prompt using model defaults, first successful output, a 5-second duration target, and the resolution closest to 720p. This route documentation lists 4-8s; 720p or 1080p; maximum 8s. The CSV does not expose the exact per-prompt overrides or criterion-specific denominators.
- Context
- Wiro AI is a commercial model gateway. Scores use its weighted 0-5 AI-judge rubric, not a blind human pairwise preference panel. No per-criterion N or confidence interval is published. This is task-specific AI-judge evidence, not broad independent quality proof.
Video Models AI-judge benchmark (Show measurement, test conditions, and source)
- Source value
- 3.62
- Score
- 3.62
- Metric
- Camera and framing AI-judge criterion score (0-5); rubric weight 12%
- Unit
- points
- Benchmark version
- Video Models leaderboard updated 2026-10-15
- Category
- video-generation
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Video Models by Wiro AI
- Status
- active
- Retrieved at
- 2026-10-16
- Methodology
- Vendor-reported result. Exact Wiro API route google/veo3-1-fast. Published method states one API run per prompt using model defaults, first successful output, a 5-second duration target, and the resolution closest to 720p. This route documentation lists 4-8s; 720p or 1080p; maximum 8s. The CSV does not expose the exact per-prompt overrides or criterion-specific denominators.
- Context
- Wiro AI is a commercial model gateway. Scores use its weighted 0-5 AI-judge rubric, not a blind human pairwise preference panel. No per-criterion N or confidence interval is published. This is task-specific AI-judge evidence, not broad independent quality proof.
Video Models AI-judge benchmark (Show measurement, test conditions, and source)
- Source value
- 4.15
- Score
- 4.15
- Metric
- Quality and detail AI-judge criterion score (0-5); rubric weight 11%
- Unit
- points
- Benchmark version
- Video Models leaderboard updated 2026-10-15
- Category
- video-generation
- Direction
- Higher is better
- Source type
- vendor-reported
- Evaluator
- Video Models by Wiro AI
- Status
- active
- Retrieved at
- 2026-10-16
- Methodology
- Vendor-reported result. Exact Wiro API route google/veo3-1-fast. Published method states one API run per prompt using model defaults, first successful output, a 5-second duration target, and the resolution closest to 720p. This route documentation lists 4-8s; 720p or 1080p; maximum 8s. The CSV does not expose the exact per-prompt overrides or criterion-specific denominators.
- Context
- Wiro AI is a commercial model gateway. Scores use its weighted 0-5 AI-judge rubric, not a blind human pairwise preference panel. No per-criterion N or confidence interval is published. This is task-specific AI-judge evidence, not broad independent quality proof.
Recorded measurement details
LMArena Text to Video: LMArena Text to Video (Show measurement, test conditions, and source)
- Source value
- 1362
- Score
- 1362
- Metric
- elo-rating
- Unit
- Elo
- Task
- lmarena-text-to-video
- Task label
- LMArena Text to Video
- Direction
- Higher is better
- Source rank
- 14
- Sample
- 39484
- Lower confidence bound
- 1352
- Upper confidence bound
- 1372
- Retrieved at
- 2026-11-05
LMArena Image to Video: LMArena Image to Video (Show measurement, test conditions, and source)
- Source value
- 1385
- Score
- 1385
- Metric
- elo-rating
- Unit
- Elo
- Task
- lmarena-image-to-video
- Task label
- LMArena Image to Video
- Direction
- Higher is better
- Source rank
- 15
- Sample
- 99804
- Lower confidence bound
- 1376
- Upper confidence bound
- 1394
- Retrieved at
- 2026-11-05
Page 1 of 9
Video sample from our test
This video exists as a documented run for the exact model version. Playback is muted.
Motion and physical consistency
This sample tests Anatomy, object consistency, camera tracking, rain, and reflections.
Prompt and settings
Generate exactly one continuous five-second 16:9 cinematic video. A fictional bicycle courier in a mustard-yellow raincoat rides a red city bicycle from left to right across a rain-wet Hamburg street at blue hour for the entire clip. The camera tracks her smoothly at waist height. Both hands remain visible on the handlebars, the bicycle geometry stays physically correct, and a transparent umbrella fixed to the bicycle moves naturally in the wind. Wet-street reflections and raindrops follow realistic physics. No cuts, no slow motion, no text, no logos, no subtitles, and no audio.
Generated through fal.ai with model fal-ai/veo3.1/fast. Archived settings were 16:9, one request, first output, requested 6 s, seed provider random, request {"prompt":"Generate exactly one continuous five-second 16:9 cinematic video. A fictional bicycle courier in a mustard-yellow raincoat rides a red city bicycle from left to right across a rain-wet Hamburg street at blue hour for the entire clip. The camera tracks her smoothly at waist height. Both hands remain visible on the handlebars, the bicycle geometry stays physically correct, and a transparent umbrella fixed to the bicycle moves naturally in the wind. Wet-street reflections and raindrops follow realistic physics. No cuts, no slow motion, no text, no logos, no subtitles, and no audio.","aspect_ratio":"16:9","duration":"6s","resolution":"720p","generate_audio":false}, configured price estimate $0.60.
Popularity compared
How often the model is searched for, used and mentioned, compared with every model of its category.
Popularity index
0 to 100, higher is more popular
20 of 98 models
As of November 2026 · Source: gradually.ai popularity index
Over time
Weekly values from October 18, 2026 to November 5, 2026
Veo 3.1 FastWan 2.7Dreamina Seedance 2.0 720pHunyuanVideo 1.5Gemini Omni 1.1 Flash
Veo 3.1 Fast compared
Compare with another model
The most popular comparisons
- 1Gemini Omni Flash vs. Veo 3.1 Fast
- 2FLUX.3 Video vs. Veo 3.1 Fast
- 3Dreamina Seedance 2.5 720p vs. Veo 3.1 Fast
- 4Dreamina Seedance 2.0 720p vs. Veo 3.1 Fast
- 5MiniMax H3 vs. Veo 3.1 Fast
- 6Sora 2 Pro vs. Veo 3.1 Fast
- 7Sora 2 vs. Veo 3.1 Fast
- 8Veo 3.1 vs. Veo 3.1 Fast
- 9Veo 3.1 Fast vs. Grok Imagine Video 1.5
Show all 48 comparisons
- 10Veo 3.1 Fast vs. Grok Imagine Video
- 11Veo 3.1 Fast vs. Wan 2.7
- 12Veo 3.1 Fast vs. Wan 2.6
- 13Veo 3.1 Fast vs. Runway Gen-4.5
- 14Veo 3.1 Fast vs. Kling 3.0 Pro
- 15Veo 3.1 Fast vs. Kling 2.6 Pro
- 16Veo 3.1 Fast vs. Vidu Q3 Pro
- 17Veo 3.1 Fast vs. PixVerse V5.6
- 18Veo 3.1 Fast vs. Hailuo 2.3
- 19Veo 3.1 Fast vs. LTX-2 19B
- 20Veo 3.1 Fast vs. Wan 2.2 A14B
- 21Veo 3.1 Fast vs. HunyuanVideo 1.5
- 22Veo 3.1 Fast vs. Pika 2.2
- 23Veo 3.1 Fast vs. Gemini Omni 1.1 Flash
- 24Veo 3.1 Fast vs. Muse Video
- 25Veo 3.1 Fast vs. Wan 3.0
- 26Veo 3.1 Fast vs. Wan 3.0 Prime
- 27Veo 3.1 Fast vs. HappyHorse 1.0
- 28Veo 3.1 Fast vs. HappyHorse 1.1
- 29Veo 3.1 Fast vs. PixVerse V6
- 30Veo 3.1 Fast vs. PixVerse C1
- 31Veo 3.1 Fast vs. Seedance 2.0 Fast
- 32Veo 3.1 Fast vs. Seedance 2.0 Mini
- 33Veo 3.1 Fast vs. Vidu Q3 Turbo
- 34Veo 3.1 Fast vs. Vidu Q2
- 35Veo 3.1 Fast vs. Vidu Q1
- 36Veo 3.1 Fast vs. Kling 4.0
- 37Veo 3.1 Fast vs. Kling 3.0 Standard API
- 38Veo 3.1 Fast vs. Kling Video O1
- 39Veo 3.1 Fast vs. Hailuo 2.3 Fast
- 40Veo 3.1 Fast vs. Wan 2.5 Preview
- 41Veo 3.1 Fast vs. Wan 2.6 I2V Flash
- 42Veo 3.1 Fast vs. Wan 2.2 TI2V 5B
- 43Veo 3.1 Fast vs. Wan 2.2 I2V A14B
- 44Veo 3.1 Fast vs. HunyuanVideo 1.0 T2V
- 45Veo 3.1 Fast vs. HunyuanVideo I2V 1.0
- 46Veo 3.1 Fast vs. Ray 3.2
- 47Veo 3.1 Fast vs. Grok Imagine Video 1.5 Lite
- 48Veo 3.1 Fast vs. LTX-2.5
All specifications
Input and output
Watermarking and provenance
AI watermarking refers to invisible signals embedded in the output. Visible marks and provenance metadata are listed separately. Unknown means there is insufficient evidence for that particular field.
- AI watermarking
- Subject to limitations. Google documents SynthID for videos from the former Gemini API preview endpoints. Those endpoints were shut down on October 22, 2026. Other access routes require separate verification.
- Source: AI watermarking
Licence and access
- Lifecycle
- Generally available on Vertex AI as veo-3.1-fast-generate-001 (retirement no earlier than November 17, 2026); the Gemini API preview veo-3.1-fast-generate-preview was deprecated by Google for October 22, 2026
- Source: Lifecycle
Prices
Free to use at Google (Google AI Free, with limited credits), more from €4.99 a month (Google AI Plus, with limited credits). Through other providers from ≈ €6.36 a month (ElevenLabs Starter, with limited credits). 58 plans in Germany.
| Plan | Monthly price | Note |
|---|---|---|
| Google AI Free | free | developer's own plan, with limited credit |
| Google AI Plus | €4.99 | developer's own plan, with limited credit |
| Google AI Pro | €21.99 | developer's own plan, with limited credit |
| Google AI Ultra 5x | €99.99 | developer's own plan, with limited credit |
| Google AI Ultra 20x | €219.99 | developer's own plan, with limited credit |
| Lovable Free | free | with limited credit |
| ElevenLabs Starter | ≈ €6.36 | with limited credit |
| Poe Basic | €6.83 | with limited credit |
| Krea Basic | ≈ €9.54 | with limited credit |
| Pika Starter | ≈ €10.60 | with limited credit |
| Adobe Firefly Standard | €10.98 | with limited credit |
| Hailuo AI StandardMiniMax | ≈ €11.13 | with limited credit |
| Leonardo.Ai Essential | ≈ €12.72 | |
| Magnific Premium | €19.04 | with limited credit |
| Recraft Pro | ≈ €21.20 | with limited credit |
| Adobe Firefly Pro | €21.99 | with limited credit |
| ElevenLabs Creator | ≈ €23.33 | with limited credit |
| Recraft Teams | ≈ €23.33 | with limited credit, for teams, per seat |
| Lovable Pro | €25 | with limited credit |
| Leonardo.Ai Team Starter | ≈ €25.45 | for teams, per seat |
| Poe Plus | €27.36 | with limited credit |
| Leonardo.Ai Premium | ≈ €31.81 | |
| Krea Pro | ≈ €37.11 | with limited credit |
| Pika Creator | ≈ €37.11 | with limited credit |
| Hailuo AI ProMiniMax | ≈ €40.29 | with limited credit |
| Magnific Premium+ | €42.84 | with limited credit |
| Lovable Business | €50 | with limited credit, for teams |
| Leonardo.Ai Team Growth | ≈ €50.89 | for teams, per seat |
| Adobe Firefly Pro Plus | €54.99 | with limited credit |
| Leonardo.Ai Ultimate | ≈ €63.61 | |
| Poe Pro | €68.41 | with limited credit |
| Higgsfield Plus | €70.21 | with limited credit |
| Magnific Business | €70.21 | with limited credit, for teams, per seat |
| Higgsfield Team | €94.01 | with limited credit, for teams, per seat |
| Hailuo AI MasterMiniMax | ≈ €94.36 | with limited credit |
| Pika Fancy | ≈ €100.72 | with limited credit |
| Magnific Pro Starter | €104.72 | with limited credit |
| ElevenLabs Pro | ≈ €104.96 | with limited credit |
| Krea Max | ≈ €111.32 | with limited credit |
| Poe Advanced | €136.84 | with limited credit |
| Higgsfield Ultra | €153.51 | with limited credit |
| Higgsfield Scale | €201.11 | with limited credit, for teams, per seat |
| Krea Business | ≈ €212.05 | with limited credit, for teams |
| Adobe Firefly Premium | €219.98 | with limited credit |
| Hailuo AI MaxMiniMax | ≈ €229.01 | with limited credit |
| Magnific Pro | €273.70 | with limited credit |
| ElevenLabs Scale | ≈ €317.01 | with limited credit, for teams |
| Poe Max | €342.11 | with limited credit |
| ElevenLabs Business | ≈ €1,049.63 | with limited credit, for teams |
| Adobe Firefly Premium für Teams | n/a | with limited credit, for teams, per seat |
| Adobe Firefly Pro für Teams | n/a | with limited credit, for teams, per seat |
| Adobe Firefly Pro Plus für Teams | n/a | with limited credit, for teams, per seat |
| ElevenLabs Enterprise | n/a | for enterprises |
| Krea Enterprise | n/a | for enterprises |
| Leonardo.Ai Custom | n/a | for enterprises |
| Lovable Enterprise | n/a | for enterprises |
| Magnific Enterprise | n/a | for enterprises |
| Pika Enterprise | n/a | for enterprises |
Frequently asked questions about Veo 3.1 Fast
How good is Veo 3.1 Fast?
Rank 11 of 23 in the text-to-video arena rating. With a start image: rank 11 of 21.
LMArena, as of Sep 22, 2026
How much does a video with Veo 3.1 Fast cost?
About $0.50 for a 5 s clip, $0.10 per second of video. That is fairly cheap.
How long can a video with Veo 3.1 Fast be?
Up to 8 seconds in one go. Resolution: 1080p.
According to Google
- Price
- $0.10 per second$0.50 per 5 s clip
- Longest clip
- 8s
- Resolution
- 1080p
- Audio
- yes, generated natively
- Open weights
- no
Where can I use Veo 3.1 Fast?
Free to use at Google. Including other providers that makes 58 plans in Germany.
Subscription comparison, prices in Germany incl. VAT
At Google
5 plans, free (Google AI Free).
Through other providers
53 plans from 11 providers, from ≈ €6.36 a month (ElevenLabs Starter), with limited credits.
Sources and data date
Show all 40 sources
Benchmark sources at a glance
Which independent evaluators tested this model. This is coverage, not a quality ranking.
| Evaluator | Test series | Composite indices | Recorded active measurements |
|---|---|---|---|
| Artificial Analysis | 0 | 0 | 4 |
| LMArena | 0 | 0 | 4 |
Without vendor claims and historical or disputed results; those stay labelled in the single values. 8 measurements are not yet assigned to a test series.
| Type | Evidence and data date |
|---|---|
| Research status | Research date October 16, 2026. 29 source URLs checked. This documents the inspected sources, not an exhaustive inventory of every publication. |
| Not found in the inspected sources | arena-video-edit-exact-row Note The inspected Video Edit table has no exact model-version row. T2V and I2V values were not reused. |
| Not found in the inspected sources | vabench-exact-model-row Note VABench has no exact model row. Sora 2 was not transferred to Sora 2 Pro or other models. |
| Unresolved | vbench-current-public-leaderboard Note The current VBench leaderboard is embedded in a dynamic Hugging Face iframe and was not retrievable; current scores remain unresolved. |
| Unresolved | vidbenchmark-performance-axes Note The source publishes speed-to-first-frame and cost-efficiency axes, but its inspected methodology does not define them as measured performance results or disclose units, direction, or a scoring formula. They are omitted. Model limits, maximum resolution, and list prices are not quality measurements. |
| Not published | AA output resolution Note The AA-Video-I2V leaderboard row does not publish the tested output resolution. |
| Not published | per-cell-confidence-intervals Original wording Tables 1, 2, 4, 8, and 9 give point scores without per-cell confidence intervals; no interval or rater-dispersion value is inferred. |
| Not published | table4-scenario-subgroup-sample-counts Original wording Table 4 does not publish sample counts for its regular, novel, and compositional scenario subgroups. The overall study sample is not assigned to each subgroup, and no subgroup N is inferred. |
| Not published | provider-route-revision-api-and-generation-seed Original wording The paper gives the exact Veo-3.1-Fast label and says it follows official/default implementations, but does not identify a provider serving route, API or revision, or generation seed. |
| Additional source | Google · Google Developers Veo 3.1 announcement · Editorial description reviewed October 2, 2026 |
| Additional source | Video pricing |
| Additional source | Lifecycle data |
| Additional source | fal.ai |
| Additional source | LMArena (retrieved November 5, 2026)12 |
| Additional source | Artificial Analysis AA-Video-T2V v2.1 (retrieved October 16, 2026; October 17, 2026; October 28, 2026) |
| Additional source | Artificial Analysis AA-Video-I2V v1.1 (retrieved October 16, 2026; October 17, 2026; October 28, 2026) |
| Additional source | Physics-IQ Verified GitHub leaderboard (retrieved October 16, 2026) |
| Additional source | Video Models by Wiro AI (retrieved October 16, 2026; October 28, 2026) |
| Additional source | vidbenchmark (retrieved October 16, 2026) |
| Additional source | SandBase API video benchmark, model/prompt score table (retrieved October 17, 2026) |
| Additional source | AVGen-Bench, Table 2 (retrieved October 17, 2026) |
| Additional source | Arena Text-to-Video leaderboard (retrieved September 21, 2026; October 16, 2026; October 17, 2026; October 28, 2026) |
| Additional source | Arena Image-to-Video leaderboard (retrieved September 21, 2026; October 16, 2026; October 17, 2026; October 28, 2026) |
| Additional source | VIBE AI Video Model Benchmarks (retrieved October 17, 2026) |
| Additional source | AVGen-Bench paper (retrieved October 17, 2026) |
| Additional source | Sci-VBench paper, arXiv:2608.09873v1 (retrieved October 17, 2026) |
| Additional source | Physics-IQ benchmark repository, merged pull request #85 (retrieved October 17, 2026) |
| Additional source | VideoASMR-Bench paper, arXiv:2512.13281v4 (retrieved October 17, 2026) |
| Additional source | OSCBench, ACL 2026 paper in ACL Anthology (retrieved October 18, 2026) |
| Additional source | RoboTrustBench (retrieved October 23, 2026) |
| Additional source | Rapidata I2V Physics benchmark (retrieved October 22, 2026; October 28, 2026) |
| Additional source | Model metadata (retrieved October 16, 2026)1234 |
| Additional source | Model metadata (retrieved October 16, 2026; October 28, 2026) |
| Additional source | Model metadata (retrieved October 17, 2026)123 |
| Additional source | Model metadata (retrieved October 28, 2026) |
| Additional source | Gemini API Veo guide (retrieved November 5, 2026) |
| Additional source | Google Gemini API deprecations · Editorial description reviewed October 2, 2026 |
| Additional source | OpenRouter: Popularity index (derived), CC BY 4.0 |
| Additional source | Vercel AI Gateway: Popularity index (derived), CC BY 4.0 |
| Additional source | Ollama: cumulative downloads for the popularity index |
| Additional source | Replicate: cumulative runs for the popularity index |
| Additional source | Google Trending Now: limited feed selection for Germany and the US |
| Additional source | Creative Commons Attribution 4.0 |