Runway Gen-4 Image Review: Legacy Gemini Benchmark
Legacy methodology: This article uses VibeDex's retired Gemini 3 Pro 200-prompt image benchmark. Its scores are historical and should not be compared directly with the Sonnet 4.6 public leaderboard. See the current Sonnet leaderboard
TL;DR
Runway Gen-4 Image is a legacy Gemini-benchmark review, not a current Sonnet recommendation. In the retired Gemini 3 Pro 200-prompt run, Runway ranked 16th of 18 at 4.064. It is not included in the current Sonnet 4.6 public image leaderboard, so the score should be read as historical context only.
Recommended Benchmarks
- Runway Gen-4 vs Hunyuan 3.0: Legacy Gemini ComparisonRunway Gen-4 Image narrowly beats Hunyuan Image 3.0 in our image benchmark, but both rank near the bottom of 18 models. See the full head-to-head breakdown.
- Hunyuan Image 3.0 Review: Legacy Gemini BenchmarkHunyuan Image 3.0 ranked 17th of 18 in our 200-prompt image benchmark at $0.080. See the full score breakdown and where it falls short on quality and value.
- Best AI Image Generator 2026: 18 Models RankedGPT Image 2 High leads our 18-model blind benchmark (4.16/5, 150 judgments per model). Full ranking, dimension winners, real costs, and where they all fail.
- Runway Gen-4.5 Review: Worth $1.21/Video? (2026)Runway Gen-4.5 scores 3.78/5 in our 10-model video benchmark. Strong Adobe workflow integration, but at $1.21/video it trails cheaper models. UPDATE 2026-05-13: independent trust audit flags Runway Auto-Fail (Trustpilot 1.10 avg, 94.7% 1-star, 44% billing complaints) — read the trust callout before signing up.
Legacy Gemini Position
The table below shows Runway's local neighborhood in the retired Gemini benchmark. It landed near the bottom of that historical run despite premium pricing.
| # | Model | Gemini Score | Cost/Image | Tier |
|---|---|---|---|---|
| 15 | Flux Dev | 4.17 | $0.003 | Budget |
| 16 | Runway Gen-4 Image | 4.06 | $0.080 | Premium |
| 17 | Hunyuan Image 3.0 | 4.04 | $0.080 | Premium |
What the Legacy Result Means
Runway remains important as a creative AI platform and video tool, but this historical still image result was weak relative to its price. That observation belongs to the retired Gemini dataset, not the current Sonnet leaderboard.
For current image-model recommendations, use the Sonnet 4.6 leaderboard. Runway Gen-4 Image does not have a current public Sonnet score.
Strengths and Limitations
Runway Gen-4 Image
Strengths
- +May matter for users already inside the Runway ecosystem
- +Historical page preserves the retired benchmark context
Limitations
- −Not present in the current Sonnet public leaderboard
- −Ranked near the bottom of the retired Gemini benchmark
- −Legacy still-image result should not be used as a current recommendation
Related Vibedex Benchmarks
GPT Image 2 vs Nano Banana Pro: Premium Benchmark
GPT Image 2 high beats Nano Banana Pro on the current Sonnet 4.6 benchmark, 4.16 to 3.96, with a 36/8/6 prompt split. GPT Image 2 medium also beats Nano Banana Pro at a much lower price.
BenchmarksBest Premium AI Image Generator 2026: Is Expensive Worth It?
GPT Image 2 High leads premium-priced public image models at 4.16. GPT Image 2 Medium and Nano Banana 2 are the practical value picks above $0.05/image.
Model ReviewHunyuan Image 3.0 Review: Legacy Gemini Benchmark
Hunyuan Image 3.0 ranked 17th of 18 in our 200-prompt image benchmark at $0.080. See the full score breakdown and where it falls short on quality and value.
FAQ
Is Runway Gen-4 Image in the current Sonnet leaderboard?
No. Runway Gen-4 Image is not present in the current Sonnet 4.6 public image dataset.
What benchmark does this review use?
This review uses the retired Gemini 3 Pro 200-prompt image benchmark, where Runway ranked 16th of 18.
Should I compare this score with current Sonnet scores?
No. Treat this as historical Gemini data only. Use the live leaderboard for current Sonnet-ranked models.
See how every model stacks up
The Vibedex leaderboard ranks 18 image models on a 50-prompt blind benchmark, judged by Claude Sonnet 4.6 across visual fidelity, physics, subject integrity, and instruction adherence.
See the leaderboard →