Comparison
Ray 3.2 vs Runway Gen-4.5: How to Choose Between Text-to-Video and Image-to-Video?
A feature-by-feature comparison of pricing, resolution, duration, aspect ratio, and first/last frames, with two sets of videos generated from the same prompt and parameters side by side, plus a scenario-based selection checklist.
If you want to generate lots of trial clips, keep per-clip costs down, or need 1080p, 21:9, or first/last frames, choose Ray 3.2. If you want fixed 720p, per-second billing, and the choice of 5/8/10 seconds, choose Runway Gen-4.5.
Ray 3.2 is now live on HiAPI. Open the text-to-video model page to generate directly. New users get 200 credits on sign-up.
See the differences at a glance
| Ray 3.2 | Runway Gen-4.5 | |
|---|---|---|
| Developer | Luma AI | Runway |
| Model ID | ray-3.2/text-to-video, ray-3.2/image-to-video |
runway-gen-4.5/text-to-video, runway-gen-4.5/image-to-video |
| Billing | Per clip: 5-second 540p starts at $0.2145 | Per second: $0.1715/second; 5 seconds costs $0.8575 |
| Resolution | 540p / 720p / 1080p | Fixed 720p |
| Duration | Text-to-video: 5 / 10 seconds; image-to-video: fixed 5 seconds | 5 / 8 / 10 seconds; text-to-video in 9:16 has no 8-second option |
| Aspect ratio | Text-to-video: 6 options, including 21:9; image-to-video follows the first frame | Text-to-video: 16:9, 9:16; image-to-video: 5 options, including 1:1, 4:3, 3:4 |
| First/last frames | Image-to-video supports a first frame plus an optional last frame | Image-to-video only takes one first frame; no last frame support |
| Prompt limit | 6000 characters | 1000 characters |
| Audio | Silent MP4 | Silent MP4 |
The 5-second 720p and 10-second 720p tiers line up directly: for 5 seconds, Ray 3.2 costs $0.429 per clip versus Runway's $0.8575; for 10 seconds, $1.286 versus $1.715. The 5-second tier comes out to exactly half.
The two handle resolution differently: Runway is fixed at 720p and the price depends only on duration; Ray 3.2 also treats resolution as a price tier, with 540p to cut costs and 1080p to gain clarity. When comparing across tiers, don't compare Ray's 540p against Runway's 720p — the unit price and visual clarity are not in the same tier.
The same prompt, one generation from each side
Both sets use the same parameters: 5 seconds, 720p, 16:9, one generation each.
Lighthouse at dusk
| Ray 3.2 | Runway Gen-4.5 |
|---|---|
Same prompt, same parameters (5 seconds, 720p, 16:9), one generation each.
Both place the camera on the spiral staircase, but the framing differs. Ray 3.2 starts outside the tower, with the camera climbing the exterior spiral staircase all the way up to the lamp room at the top; the keeper is a small figure on the stairs, waves break white around the base of the tower, and the image leans cool blue. Runway Gen-4.5 puts the camera inside the tower, following a keeper in a dark coat up the interior spiral staircase, with the sea visible through arched windows; the figure occupies the foreground for the first half, and the image leans gray-green. Both show the keeper, the spiral staircase, and the stormy sea; Ray's endpoint is the lamp room, while Runway's endpoint is the figure's back.
A lone lighthouse keeper climbs the spiral stairs at dusk, storm waves crash against the rocks below, the beam sweeps across the sea, slow cinematic crane shot
Sunrise rice terraces
| Ray 3.2 | Runway Gen-4.5 |
|---|---|
Same prompt, same parameters (5 seconds, 720p, 16:9), one generation each.
Both preserve golden hour, flooded terraces, and a single farmer. Ray 3.2's camera is higher and closer to a top-down view; the terrace ridges fill the frame, the farmer is just a dot on the path, and the sun is off-screen most of the time, leaving only a warm glow in the upper left, while the camera drifts laterally at high altitude. Runway Gen-4.5's camera is lower, with both the horizon and the sun in frame; the farmer, wearing a conical hat and pushing a cart along the terrace ridge, occupies the main position in the lower half of the frame, with a treeline and thin mist visible in the distance, while the camera tracks laterally with the farmer.
Aerial establishing shot drifting slowly over a terraced rice valley at golden hour, a low sun raking across flooded paddies that mirror an amber sky. Thin mist pools in the lower terraces. The camera moves laterally with a slow, patient pace, gradually revealing a single farmer walking a narrow dirt path. Warm high-dynamic-range lighting, rich greens and golds, cinematic haze. Negative: no text, no watermark, no flicker, no distorted geometry.
How should the two workflows be paired?
Compare text-to-video with text-to-video and image-to-video with image-to-video; don't compare one side's text-to-video with the other side's image-to-video. Runway Gen-4.5 image-to-video takes one first frame via image_urls, suited to existing footage where you only want to add motion; Ray 3.2 image-to-video uses image_url and can also add end_image_url to lock the ending, suited to shots with a clear endpoint such as blooming flowers, sunrise, or before/after transformations.
Durations don't map one-to-one either: Ray 3.2 text-to-video only offers 5 and 10 seconds, while Runway Gen-4.5 adds 8 seconds; image-to-video is fixed at 5 seconds on one side, while the other offers 5/8/10 seconds. Choose the workflow based on the duration the platform for your final cut requires, then compare unit prices.
How to choose by scenario?
- Batch vertical short videos, cost-sensitive per clip → Ray 3.2; a 5-second 540p clip costs $0.2145, so $10 gets you about 46 clips.
- Need a 1080p final cut → Ray 3.2; Runway Gen-4.5 is fixed at 720p.
- Text-to-video in 21:9, 1:1, 4:3, or 3:4 → Ray 3.2; Runway text-to-video only has 16:9 and 9:16.
- Have an image to animate and want to lock the ending frame → Ray 3.2 image-to-video; pass the last frame via
end_image_urlat no extra cost. - Need the intermediate 8-second length, or prefer to calculate cost by the second → Runway Gen-4.5.
- Very long prompts (brand guidelines or a full storyboard script pasted into the request) → Ray 3.2 supports 6000 characters; Runway caps at 1000 characters.
- Need the final video to include audio → both output silent MP4s; add the audio track during editing.
Is switching from one to the other a hassle?
No. Both models use POST https://api.hiapi.ai/v1/tasks, the same API Key, and both return a taskId first, then you poll GET /v1/tasks/:id or receive a callback.url. Switching only requires changing three things:
model: changeray-3.2/text-to-videotorunway-gen-4.5/text-to-video.duration: Ray 3.2 only accepts the integers 5 or 10; Runway accepts 5, 8, 10 (no 8 seconds for 9:16). Both are numbers, so no string conversion is needed.- Image-to-video fields: Ray 3.2 uses
image_urlplus optionalend_image_url; Runway uses theimage_urlsarray with a single image and doesn't support a last frame.
prompt and aspect_ratio are written the same way. resolution is only accepted by Ray 3.2; Runway is fixed at 720p, so you don't need to pass it.
For a project already running on one model, keep the prompt unchanged, switch only model, and run one clip to confirm the fields and billing are correct before migrating in bulk.
FAQ
How much cheaper is Ray 3.2 than Runway Gen-4.5? 5-second 540p is 75% cheaper than Runway, and 5-second 720p is 50% cheaper; 1080p flips the comparison: 5-second 1080p on Ray 3.2 costs twice as much as Runway, and 10-second 1080p costs three times as much.
Can Runway Gen-4.5 output 1080p? No, both workflows are fixed at 720p.
Can Ray 3.2 image-to-video add a last frame?
Yes. Pass the first frame via image_url and the last frame via end_image_url, at no change in price.
Can both output vertical video? Yes. Ray 3.2 text-to-video lets you choose 9:16 directly; Runway text-to-video also offers 9:16, with a choice of 5 or 10 seconds.
Is Ray 3.2's 10-second option more expensive than Runway's? 540p and 720p are cheaper, while 1080p is more expensive: for 10 seconds, 540p is $0.643 versus $1.715, 720p is $1.286 versus $1.715, and 1080p is $5.143.
Start by running the same lighthouse prompt on Ray 3.2 as a 5-second 720p clip ($0.429). Open the model page with the prompt already filled in →
Ray 3.2 series articles
- Ray 3.2 text-to-video model page: generate online and see real examples
- Ray 3.2 image-to-video model page: first-frame, last-frame, and vertical video examples
- How much does the Ray 3.2 API cost: 6 pricing tiers and monthly budgets
- How to use the Ray 3.2 API: get your first video running in 5 minutes
- Ray 3.2 prompt guide: 5 text-to-video + 3 image-to-video examples
For Runway Gen-4.5's unit prices and integration: How much does the Runway Gen-4.5 API cost, How to use the Runway Gen-4.5 API.
Related articles
View all articles →-
Comparison
Recraft v4.1 Flash Text-to-Image Horizontal Comparison: Developer Model Selection and Migration Guide
Use the $0.01 per-image cost and pure text-to-image capability to judge whether it fits your image generation pipeline
-
Comparison
hailuo-2.3-fast/image-to-video Side-by-Side Comparison: How to Choose an Image-to-Video Model and Key Migration Points
-
Comparison
Seedream 5.0 Pro Image-to-Image Side-by-Side Comparison: Capabilities, Pricing, and Selection Guidance