Tired of ads? Enjoy an ad-free experience by signing up.
👩‍🍳 How we use AI at Tech in Asia, thoughtfully and responsibly.
🧔‍♂️ A friendly human may check it before it goes live. More news here

US AI video startup Runway debuts new Nvidia-powered model

Runway has announced gen-4.5, the latest version of its AI-powered video generation model, on December 1.

The company said gen-4.5 improves motion quality, prompt accuracy, and visual detail, while maintaining the speed and efficiency of its predecessor.

Runway claims gen-4.5 leads the Artificial Analysis Text to Video benchmark with 1,247 Elo points.

The model supports various generation modes, including image-to-video and keyframes, and runs on Nvidia Hopper and Blackwell GPUs.

🔗 Source: Runway

🧠 Food for thought

Implications, context, and why it matters.

Independent benchmarks back Runway’s lead amid limited third-party testing

  • Runway’s gen-4.5 ranks first on Artificial Analysis’s Video Arena (a third-party blind head-to-head leaderboard) 1. Voters compare clips without knowing the maker, placing it above Google’s Veo 3 (a competing AI video model) in second and OpenAI’s Sora 2 Pro (another leading AI video generator) in seventh.
  • Crowdsourced votes on the arena offer an outside check on performance that goes beyond Runway’s own claims 1.
  • Broad third-party shootouts that pit gen-4.5 against top rivals across varied prompts and failure modes (tests for specific error cases) are rare, which leaves questions about consistency outside benchmark settings.

GPU clouds aim to meet video inference demand with prebuilt Blackwell instances

  • Gen-4.5 runs on NVIDIA Hopper and Blackwell GPUs, so GPU clouds such as Nebius (a GPU-focused cloud platform) already sell self-serve HGX B200 capacity (NVIDIA’s Blackwell-based server platform) with pay-as-you-go pricing, no waitlist 2.
  • Server makers plan NVIDIA RTX PRO Servers with RTX PRO 6000 Blackwell GPUs (workstation/data center GPUs based on the Blackwell generation) 3. Dell Technologies and Hewlett Packard Enterprise (HPE) join Lenovo plus Supermicro, with some SKUs shipping now, 2U (two rack units high) mainstream systems due later this year.
  • Cloud orchestrators (cloud platforms plus managed service providers that automate provisioning and deployment) can reserve Blackwell capacity now, then ship pre-configured containers with tuned drivers for video generation, which sets them up to meet enterprise demand through 2025.

Stay ahead in Asia’s tech landscape

You've reached your 2 free content limit for the month. Sign up for free to read the full story.

🏄 For casual readers / 👶 Free

Basic

US$0

Free forever

Get instant access to this article and more every month

0 premium content

Unlimited news briefs

5

5 articles

Ad-free reading experience

Just US$0 per day

⌛Sign up in 20s. No payment details needed.

📖 For learners / 👍 Starter

Lite

US$4.92/month

Billed annually at US$59/year

Get instant access to this article and more every month

4

4 premium content

Unlimited news briefs & articles

Ad-free reading experience

Just US$0.17 per day

Cancel anytime

Our subscriber community includes professionals from these companies:

Stay updated on the go with our mobile app.

Get latest insights with smoother, more personalized experience through TIA mobile app.