
15 Sep 2026
Creatify Labs launches Boreal, a cent-per-second realtime video model
Creatify Labs said it launched Boreal, an AI video model for text-to-video and image-to-video billed at about one cent per second of finished video and aimed at realtime speed — roughly a five-second clip in about five seconds. The company says blind human reviews and ad-quality checks put Boreal near top closed models advertisers use today, at far lower cost and about 40 times the speed in its benchmarks.
Same-day SOFTWARE primary that an ad-focused open-base video model is shipping as a priced, realtime API at roughly a cent per second, with the company publishing blind-review numbers against both its base weights and a closed rival.
Creatify Labs announced general availability of Boreal on 15 Sep 2026. Boreal, as the company tells it, is an AI video model for text-to-video and image-to-video — type a prompt, or start from a still picture, and get a clip. The company PRIMARY via PRWeb / PR Newswire is dated September 15, 2026, 10:00 ET, and datelined Mountain View, Calif. Creatify says the model is available now through the Creatify Labs API, Creatify’s Model Playground, AdFlow, and the fal platform. Fal here is a third-party host that lets developers call AI models over an API. Those company pages are the filing event. These are company claims. This desk did not run Boreal.
Pricing, still company: billed per second of finished video at about one cent a second — about $0.05 for a five-second clip — rather than per credit or token, so the bill can be worked out before a job runs. Creatify says that is 6 to 11 times cheaper per clip than the models in its benchmark set. File the cent-per-second price, the five-second about-$0.05 line, and the 6–11× cheaper claim as Creatify’s. This desk did not buy a second of video or audit other vendors’ list prices.
Speed, as the same wire has it: the model serves at about 1:1 realtime — a second of finished video for every second of GPU time — versus the minutes-long queues on closed models it measured. Closed models here means video generators whose weights are not published. Creatify says that is about 40 times the speed of a leading closed model on identical inputs. File the 1:1 realtime line and the ~40× speed claim as Creatify’s. This desk did not time a render.
Base, still company: Boreal is built on Lightricks LTX-2.5, an open-source video model with 22 billion parameters that generates audio and video together. Creatify says it applied one post-training recipe — extra training on advertising footage after the base model was already trained — and kept the rest identical. File LTX-2.5, the 22-billion-parameter joint audio-and-video base, and that ad-footage post-training as Creatify’s. This desk did not inspect the weights.
Quality claims, company report — attribute, do not desk-verify: in blind side-by-side review, evaluators preferred Boreal over the untouched base in 81% of decisive comparisons (25 to 6, with 9 ties, across 40 production cases; sign test p<0.001). A sign test here is a stats check that the preference is unlikely to be a coin flip. Against a leading closed model on identical inputs, Creatify says Boreal matches on the perceptual ad-quality measures that decide an ad and leads on motion, at roughly 40 times the speed and 12 times cheaper at the benchmarked price. On raw fidelity, the company says the figures land within a few percent either way when the closed model runs at higher resolution. File those review counts, the 81% line, the match-and-lead claim, and the few-percent fidelity caveat as Creatify’s. This desk did not rerun the comparisons or sit with the reviewers.
Product-consistency focus, still company: Creatify says general video models often break the two things an ad is bought on — they swap a product label for an invented one, morph the product, or replace the creator mid-clip. The post-training, the company says, targets keeping product labels and names and the creator’s identity stable across a clip. File that consistency pitch as Creatify’s. This desk did not watch a label hold.
Volume example and limits, company — attribute: a storyboard of twenty five-second shots costs about a quarter of a dollar and finishes in under two minutes on Boreal, versus roughly $12 and over an hour of queue time on a leading closed model, on Creatify’s measurements. Creatify is also direct about where the model stands: strongest on short clips and simple scenes; the narrowest margin over its base is on single-person talking clips. File the volume example and those limits as Creatify’s. This desk did not generate a storyboard.
Plain English for the rest of the card: text-to-video = type a prompt, get a clip. image-to-video = start from a still picture, get a clip. realtime here = generation time roughly matches clip length. LTX-2.5 = the open Lightricks base model Creatify fine-tuned. post-training = extra training on ad footage after the base model was already trained. fal = a third-party host for calling AI models over an API.
PRIMARY here: Creatify Labs’ 15 Sep 2026 PRWeb / PR Newswire company release — Tier A PRIMARY company source, the original record. The product page is product-home context, not a second originating newsroom. The Boreal GA, the text-to-video / image-to-video offer, the Creatify Labs API / Model Playground / AdFlow / fal availability, the about-one-cent-per-second / about-$0.05-for-five-seconds price, the 6–11× cheaper claim, the 1:1 realtime and ~40× speed lines, the LTX-2.5 / 22-billion-parameter / ad-footage post-training base, the 81% blind-review counts, the match-on-ad-quality / lead-on-motion / few-percent-fidelity claims, the product-and-creator consistency pitch, the twenty-shot volume example, and the short-clip / simple-scene / talking-clip limits are company-attributed. Quality, cost, and speed figures stay company-attributed. This desk did not independently re-run benchmarks. NOT claimed: that Boreal is the best video model, named customers, funding, a valuation, independently verified review or price comparisons, that this desk generated a clip, a stock tip, or investment advice. Distinct from the already-filed crusoe-perplexity-partnership, profound-180m-series-d, wso2-agent-manager-ga, and meta-one-subscription.
RELATED
- Meta launches Meta One, a paid plan for more AI usage across its apps
- WSO2 ships Agent Manager GA to govern enterprise AI agents without locking to one stack
- Profound raises $180M Series D at $1.8B to build an AI marketing platform
- Crusoe and Perplexity ink a multi-year deal for train-to-serve AI cloud