TensorScaleAI

Powering the future of generative worlds

[Yao Zhang @YaoZhangX]

I'm thrilled to welcome Yuan Tang (@TerryTangYuan) as Co-Founder of TensorScale!
I’ve known Yuan since we were colleagues at Ant Group. He’s one of the top experts in cloud-native AI infrastructure and large-scale model serving, leading projects across KServe, Kubeflow, Argo, and Kubernetes.
At TensorScale, we’re rebuilding the AI stack for world models. That means pairing fast inference with efficient compute scheduling, high GPU utilization, and reliable global serving across data centers, all critical to supporting our customers and their applications at massive scale.
Yuan brings deep experience turning complex infrastructure into production systems. I’m excited to build this next chapter together! 🚀

[Yuan (Terry) Tang @TerryTangYuan]

Excited to share that I'm starting a new chapter as Co-Founder of @TensorScaleAI!

I'm joining forces with a group of fellow builders to co-found TensorScale AI. Together, we're tackling one of the next big challenges in AI: rethinking the infrastructure needed for world models.

The AI stack was largely built for language models. As AI expands into video, generative media, robotics, and interactive worlds, we believe it's time to rethink the infrastructure from the ground up: making inference faster, more efficient, and more accessible to developers.

I'm excited to help shape this vision, build the technology, and grow a company from the ground up with an incredible team.

We're just getting started! 🚀

Learn more: https://t.co/zHPUBkEzUG

[Yao Zhang @YaoZhangX]

FLUX 3 Image Klein is now available on TensorScale!

[Yao Zhang @YaoZhangX]

We're hiring! 🚀
Join us to train fast, high-quality multimodal models and/or build agents that help creators plan, generate, critique, and refine multimodal content. Your research goes into production, where real users feel the difference.
https://t.co/s6FX6sPHck

[Yao Zhang @YaoZhangX]

When you can generate faster than playback, it unlocks entirely new experiences and product possibilities.
I’m thrilled to announce two new demos from TensorScale:
Multiverse Studio — turn a single product image into a full set of campaign content across different scenes and worlds.
Infinite TV — an endless interactive journey through generated worlds, where you choose what happens next.
Try them at https://t.co/5A6G1jRAFf and come build with us.

[TensorScaleAI @TensorScaleAI]

FastH3 v2 is now live on TensorScale!
To celebrate the launch, we’re giving new users $10 in credits for a limited time, that’s up to 250 seconds of H3 video generation.
Try it out at https://t.co/Tt4Mm9FiCC
Huge credits to @haoailab for making this high-quality, post-trained version of H3 available.

[Yao Zhang @YaoZhangX]

We’re hiring a Founding GTM Lead at TensorScale.

Since launch, we’ve seen tremendous interest from developers and AI teams. Speed, cost, and ease of integration are consistently resonating with customers. Thank you to everyone who has tried TensorScale and shared feedback with us.

What our early customers are saying:
“Your endpoints are super fast, it’s really impressive.”
“It is indeed very fast, an order of magnitude faster in fact!”
“We have already integrated it into our app, and we’ve been surprised by how fast it works! The API docs are very clear, and we didn’t have any issues integrating it.”
“I’m currently integrating TensorScale into my pipeline. The experience has been smooth, and I’m really happy with the results I’ve seen so far.”

Now we’re looking for someone to help turn this momentum into growth — owning GTM strategy, positioning, customer acquisition, distribution, conversion, and expansion.

Interested? DM me or apply here:
https://t.co/2nGBjez3Ez

[Yao Zhang @YaoZhangX]

@MiniMax_AI H3 ref-to-video is now live on @TensorScaleAI , with near-real-time inference

This brings together Alibaba Cloud PAI’s distilled H3 model with TensorScale’s inference stack, achieving near-real-time speed: 17.7 seconds to generate 15-second 768p video for a real time factor (RTF) of 1.18. We’re continuing to optimize the stack and expect to push RTF below 1, enabling faster-than-real-time generation.

Huge credit to MiniMax for the foundational model work, and to the Alibaba PAI team and broader open-source community for pushing the ecosystem forward. We’re excited to build on these efforts and make high-quality video generation faster, cheaper, and easier to serve in production.

[TensorScaleAI @TensorScaleAI]

RT @YaoZhangX: SenseNova U1.5 (@SenseTime_AI) is now live on @TensorScaleAI: frontier-quality image generation at 10x the speed and 60% of…

[Yao Zhang @YaoZhangX]

LTX 2.5 JUST WENT FASTER THAN REAL TIME

LTX (@ltx_io) is already one of the world's most widely used open video models. We just made its latest release, LTX 2.5, fast enough for rapid iteration and affordable enough for production at scale. 

It’s live on @TensorScaleAI now, generating a 5-sec 1080p video in just 3.12s at $0.04 per second. That's faster than real time, at a fraction of the cost!

For creative teams, that means less waiting time, more ideas explored, and lower production costs.

For robotics and physical AI teams, that means more data generated at scale, more scenarios tested, and shorter development and deployment cycles.

Request early access today and we’ll get you onboarded!
(link: https://t.co/0PAXBDiNir)

[Yao Zhang @YaoZhangX]

Today, we’re coming out of stealth with TensorScale (@TensorScaleAI ), and we're already running the fastest video inference in the world.
MiniMax (@MiniMax_AI ) H3 — one of the leading open video models — now runs 10x faster and at half the cost on our inference stack.
I’m building the company with long-time colleagues Teng Teng (@tt_tsio), Ye Wang (@yewang_ts), and Yun Tang (@_tangyun_). The AI stack was built for language models. We're rebuilding it for video and world models.