Palette 3.0 Hub

Any Physical AI Model, Performance that matters

Demos, white papers, blogs and webinars from the Palette 3.0 release. See what Palette does on purpose-built hardware, then go build it yourself.

PALETTE 3.0 NOW AVAILABLE

ANY Physical AI Model. REAL Machine Autonomy.

  • Agentic ISP
  • BEV kernels
  • Camera–LiDAR
  • 60 channels

3X greater perception, 15X faster camera tuning

  1. Ultra-High-Density AI Video Processing: Process up to 60 independent 720p streams on a single Modalix MLSoC or 120 streams on two PCIe HHHL Cards

  2. Agentic ISP tuning. Tune cameras without recompiling

  3. Optimized BEV kernels: Resolved primary computational bottlenecks in 3D perception pipelines.

Webinars · Join us live

Explore what’s possible with Palette 3.0

Get the technical details from the team that built it, see real use cases, and bring your questions.

Inside Palette 3.0: Any Physical AI Model. Real Machine Autonomy.

Everything new in Palette 3.0. 45 minutes plus demos and live Q&A.

Register now

Drone and Robot Intelligence at Flight Speed

What’s new for drones and robots and how to build, develop and deploy faster. 30 minutes plus demos and live Q&A.

Register now

Demos

See it run.

Watch Palette 3.0 in action on Modalix.

  • 60 Channel AI Video Processing On A Single Modalix MLSoC

    In this demo, SiMa.ai’s Modalix MLSoC is running YOLOv26 object detection across 60 simultaneous 720p video streams, each at 10 frames per second.

  • Concurrent AI Model Processing on Modalix MLSoC

    In this demonstration, four different AI models are running concurrently on a single SiMa.ai Modalix MLSoC in a project built with the Palette Software Suite.

  • Multiple Docker Containers Running On Modalix MLSoC

    SiMa.ai now supports Docker containers on the Modalix MLSoC, making it easier than ever to develop, deploy, and manage physical AI applications.

  • Ultra-High-Density AI Video Processing on Modalix MLSoC PCIe Cards

    In this demonstration, we're running AI video analytics on 120 simultaneous video channels, using the YOLOv26 object detection model on two Modalix PCIe HHHL cards.

Next step

Pick where to start.

Commitment: A conversation
Try It Yourself Complete hardware development platform. Ready to deploy your first model on day one.
Explore more See the product brief and visit our developer and community pages.

Talk to an expert

For teams who want their migration scoped.

By registering, you agree to our  Privacy Policy.