V-Benchmark v2.0 comparison

LTX 2.3 22B vs Gen-4.5

By Megaton TeamSeptember 1, 2026

Gen-4.5 leads this v-benchmark v2 comparison on the published overall score.

Share
LTX 2.3 22B evaluation example
LTX 2.3 22B
Gen-4.5 evaluation example
Gen-4.5
Our verdict

Gen-4.5 leads

Gen-4.5 leads this v-benchmark v2 comparison on the published overall score.

Lightricks

Lightricks

LTX 2.3 22B

51.8Overall score
Rank
#14
Price per minute
$2.13/min
Runway

Runway

Gen-4.5

60.5Overall score
Rank
#13
Price per minute
$7.20/min

Best for

Highest overall score

Gen-4.5

Gen-4.5 leads the overall score by 8.7 points.

The overall score does not measure file reliability.

Best for

More readable text

Gen-4.5

Gen-4.5 leads by 35.6 points.

Long or moving text still needs manual review.

Best for

Lower generation cost

LTX 2.3 22B

LTX 2.3 22B runs $2.13/min, against $7.20/min for Gen-4.5.

Provider prices can change.

Best for

Built-in audio

LTX 2.3 22B

LTX 2.3 22B generates audio with the video; Gen-4.5 does not.

This comparison does not score audio quality.

0
videos rated
0
measurements
0
ratings
September 1, 2026
last updated

Reviewer scores

Scores by category

Gen-4.5 wins 12 of 13 categories head-to-head. Each bar is the reviewer score out of 100.

Biggest gapText Fidelity: Gen-4.5 ahead by 35.6 points.
Closest categoryPhysics: LTX 2.3 22B ahead by 1.2 points.

Head-to-head scores

13 categories rated head-to-head

LTX 2.3 22BGen-4.5
Physics
40.0
38.8
Prompt Adherence
73.0
76.0
Logic Consistency
46.4
72.7
Scene Consistency
64.2
78.1
Cinematography
58.6
70.0
Taste
42.0
49.7
2D Animation
33.3
47.5
3D Animation
42.5
44.0
Human
50.7
67.4
Actor Performance
41.0
63.0
Objects
48.3
74.4
Animals
44.3
61.4
Text Fidelity
11.6
47.2

Capabilities

Inputs and outputs

CapabilityLTX 2.3 22BGen-4.5
Text inputYesYes
Image inputYesYes
Video inputYesYes
Audio inputYesNo
Image outputNoNo
Audio outputYesNo

FAQ

Common questions

Is LTX 2.3 22B better than Gen-4.5?

Not on our scores — Gen-4.5 edges ahead overall, 60.5 to 51.8, winning 12 of 13 rated categories. LTX 2.3 22B still leads in a few areas the category scores make clear.

Is LTX 2.3 22B cheaper than Gen-4.5?

Yes. LTX 2.3 22B runs $2.13/min, against $7.20/min for Gen-4.5.

Which makes longer clips, LTX 2.3 22B or Gen-4.5?

Gen-4.5 — up to 10 seconds per clip, versus 5 seconds for LTX 2.3 22B.

Does LTX 2.3 22B generate audio?

Yes — LTX 2.3 22B generates audio together with the video, in the same file.

Does Gen-4.5 generate audio?

No — Gen-4.5 outputs silent video; you'd add sound separately.

Is LTX 2.3 22B or Gen-4.5 better for text fidelity?

Gen-4.5 leads text fidelity, by 35.6 points.

Is LTX 2.3 22B or Gen-4.5 better for logic consistency?

Gen-4.5 leads logic consistency, by 26.4 points.

Is LTX 2.3 22B or Gen-4.5 better for objects?

Gen-4.5 leads objects, by 26.2 points.

Is LTX 2.3 22B or Gen-4.5 better for actor performance?

Gen-4.5 leads actor performance, by 22.0 points.

Is LTX 2.3 22B or Gen-4.5 better for animals?

Gen-4.5 leads animals, by 17.1 points.

Is LTX 2.3 22B or Gen-4.5 better for human?

Gen-4.5 leads human, by 16.7 points.

Is LTX 2.3 22B or Gen-4.5 better for 2d animation?

Gen-4.5 leads 2d animation, by 14.2 points.

Is LTX 2.3 22B or Gen-4.5 better for scene consistency?

Gen-4.5 leads scene consistency, by 13.9 points.

Is LTX 2.3 22B or Gen-4.5 better for cinematography?

Gen-4.5 leads cinematography, by 11.4 points.

Is LTX 2.3 22B or Gen-4.5 better for taste?

Gen-4.5 leads taste, by 7.7 points.

How are LTX 2.3 22B and Gen-4.5 tested?

Both models run the same prompts. Professional reviewers score each video 0–100 on how well it follows the prompt and handles motion, people, objects, animation, and text, while automated checks flag freezes, bad framing, and unstable motion. This matchup draws on 0 rated videos and 0 reviewer ratings.

More comparisons

More AI video model comparisons

Browse more head-to-head results, including scores, prices, and side-by-side video tests.

Other AI video model comparison pages
ComparisonProvidersOverall scores
Gen-4.5 vs Seedance 2.5RunwayByteDance61.0 / 87.0
Gen-4.5 vs Wan 3.0RunwayAlibaba61.0 / 82.0
LTX 2.3 22B vs Seedance 2.5LightricksByteDance52.0 / 87.0
Gen-4.5 vs Seedance 2.0RunwayByteDance61.0 / 78.0
LTX 2.3 22B vs Wan 3.0LightricksAlibaba52.0 / 82.0
FLUX 3 vs Gen-4.5Black Forest LabsRunway77.0 / 61.0
LTX 2.3 22B vs Seedance 2.0LightricksByteDance52.0 / 78.0
FLUX 3 vs LTX 2.3 22BBlack Forest LabsLightricks77.0 / 52.0
Gemini Omni 1.1 Flash vs Gen-4.5GoogleRunway76.0 / 61.0
Gemini Omni 1.1 Flash vs LTX 2.3 22BGoogleLightricks76.0 / 52.0
Kling 3 Pro vs Gen-4.5Kling AI (Kuaishou)Runway76.0 / 61.0
Kling 3 Pro vs LTX 2.3 22BKling AI (Kuaishou)Lightricks76.0 / 52.0
MiniMax Hailuo 03 vs Gen-4.5MiniMaxRunway75.0 / 61.0
LTX 2.3 22B vs MiniMax Hailuo 03LightricksMiniMax52.0 / 75.0
MiniMax H3 Max vs Gen-4.5fal.aiRunway75.0 / 61.0
Gemini Omni Flash vs Gen-4.5GoogleRunway70.0 / 61.0
MiniMax H3 Max vs LTX 2.3 22Bfal.aiLightricks75.0 / 52.0
Gemini Omni Flash vs LTX 2.3 22BGoogleLightricks70.0 / 52.0
Kling 3 Standard vs Gen-4.5Kling AI (Kuaishou)Runway69.0 / 61.0
Gen-4.5 vs Seedance 2.0 MiniRunwayByteDance61.0 / 69.0
Kling 3 Standard vs LTX 2.3 22BKling AI (Kuaishou)Lightricks69.0 / 52.0
Gen-4.5 vs Veo 3.1RunwayGoogle61.0 / 63.0
LTX 2.3 22B vs Seedance 2.0 MiniLightricksByteDance52.0 / 69.0
LTX 2.3 22B vs Veo 3.1LightricksGoogle52.0 / 63.0
Seedance 2.5 vs Wan 3.0ByteDanceAlibaba87.0 / 82.0
Seedance 2.0 vs Seedance 2.5ByteDanceByteDance78.0 / 87.0
FLUX 3 vs Seedance 2.5Black Forest LabsByteDance77.0 / 87.0
Seedance 2.0 vs Wan 3.0ByteDanceAlibaba78.0 / 82.0
FLUX 3 vs Wan 3.0Black Forest LabsAlibaba77.0 / 82.0
Gemini Omni 1.1 Flash vs Seedance 2.5GoogleByteDance76.0 / 87.0
FLUX 3 vs Seedance 2.0Black Forest LabsByteDance77.0 / 78.0
Gemini Omni 1.1 Flash vs Wan 3.0GoogleAlibaba76.0 / 82.0

Methodology

How we tested LTX 2.3 22B and Gen-4.5

We score two things separately: how good the video looks to a reviewer, and how reliably the file itself holds up. A model can make great-looking footage and still deliver clips with freezes, bad framing, or unstable motion.

We test every model with v-benchmark v2.0, our benchmark for AI video models — the same prompts, professional reviewers, and automated reliability checks behind every score here.

Reviewer scores
Both models receive the same prompts. Professional reviewers score how well each video follows the prompt and handles motion, composition, people, objects, animals, animation, and on-screen text. Scores use a 0-100 scale.
File reliability
Automated checks review 0 clips for freezes, bad framing, compression problems, unstable motion, distorted geometry, exposure problems, and invalid file properties.
Missing scores
Categories with a score for only one model are shown but excluded from the head-to-head win count. Price and generation speed stay separate because they use different units.
Last updated
Reviewer scores and rankings use results from September 1, 2026. Reliability results use the latest completed test for each model. Provider prices and rankings can change.
Subscribe to the Megaton newsletter

Get weekly AI video rankings, evaluation updates, and industry news delivered to your inbox.