Periodic Neon

AI model · Science / biology model by Periodic Labs

Accelerating scientific discovery with 1T-parameter autonomous analysis.

A 1T-parameter scientific AI model post-trained for experimental analysis and XRD tasks. It is designed for researchers in the physical sciences to accelerate autonomous discovery through lab-integrated deployment.

Model specs

The full spec card (context, pricing per token, modalities and more) on AntSpace. Sign up free to see the full spec card.

Launched
2026-09-15
Languages
English

Category: Generative AI

Parameters
1 Trillion
Released
2026-09-15
Parameter Count
1 Trillion
Training Method
Scientific midtraining and reinforcement learning
Primary Application
XRD analysis and experimental data interpretation
Deployment Environment
Physical laboratories and sandboxed infrastructure
  • 1T-parameter scientific AI model
  • Post-trained for experimental result analysis
  • Pareto-optimal for challenging XRD analysis
  • Deployed directly to physical laboratories

Overview

Periodic Neon is a 1-trillion-parameter scientific AI model developed by Periodic Labs to accelerate discovery in the physical sciences. The model is specifically post-trained to analyze experimental results and is deployed directly within the company's physical laboratories to facilitate autonomous discovery. By utilizing scientific midtraining and reinforcement learning on unique lab data, Periodic Neon achieves Pareto-optimal performance for challenging tasks such as X-ray diffraction (XRD) analysis. It is designed for researchers and engineers who need to make sense of vast amounts of experimental data to iterate faster on materials design and physical simulations. The model operates within a specialized infrastructure designed for scientific AI efficiency at scale, including dedicated training, inference, and sandboxing environments.

Scientific Analysis

  • Experimental Analysis: The model is specifically post-trained to analyze and interpret experimental results from physical laboratories.
  • XRD Analysis: Periodic Neon is the Pareto-optimal model for challenging X-ray diffraction analysis tasks.
  • Lab Deployment: The model is deployed directly into physical labs to facilitate autonomous discovery.

Training and Infrastructure

  • Scientific Midtraining: The model undergoes scientific midtraining on high-quality data generated by autonomous labs.
  • Reinforcement learning is applied using lab data to refine the model's performance in verifiable environments.
  • Sandboxing Infrastructure: The model operates within a sandboxing infrastructure designed for scientific AI efficiency at scale.

Benchmarks

  • XRD analysis: Pareto-optimal performance

Use cases

  • Scientists use Periodic Neon to analyze complex X-ray diffraction (XRD) data to identify new material structures.
  • Researchers deploy the model in autonomous labs to interpret experimental results and guide the next steps of discovery.
  • Engineers at semiconductor manufacturers use custom agents powered by the model to analyze heat dissipation data for faster chip iteration.

Specifications

Domains
Physical sciences, materials design, semiconductor manufacturing
Parameters
1 Trillion
Released
2026-09-15
Parameter Count
1 Trillion
Training Method
Scientific midtraining and reinforcement learning
Primary Application
XRD analysis and experimental data interpretation
Deployment Environment
Physical laboratories and sandboxed infrastructure

FAQ

What is Periodic Neon?

Periodic Neon is a 1T-parameter model that has been post-trained specifically to analyze experimental results and is deployed directly to physical labs.

What scientific tasks can it perform?

The model is primarily used for challenging X-ray diffraction (XRD) analysis and interpreting experimental data in the physical sciences.

How was the model trained?

It is trained using scientific midtraining and reinforcement learning on high-quality data generated by autonomous laboratories.

Who is this model for?

The model is designed to put autonomous discovery into the hands of scientists, helping them iterate faster on materials design and experimental data.

Added by AntSpace from public sources (periodic.com).

Related

More AI models by Periodic Labs