Skip to content
AI News HubLIVE
More
Summary only

Zyphra Releases ZAYA1-8B Reasoning Model

ZAYA1-8B is a mixture-of-experts language model that matches or exceeds substantially larger open-weight models on complex reasoning, mathematics, and coding tasks while using fewer than one billion active parameters. Trained on full-stack AMD infrastructure, it introduces Markovian RSA for test-time compute and is open-sourced under Apache 2.0.

SourceAIwireAuthor: Andrew Jolly
Zyphra Releases ZAYA1-8B Reasoning Model
Report an error

The correction channel is not available yet. You can copy the article reference below for later.

Correction instructions

Summary only

Under the source content permission, this page provides an editorial summary, key points, and a link to the original article.

Key points and analysis

Article intelligence

EngineersAdvanced

Key points

  • ZAYA1-8B achieves competitive performance with models many times its size using under 1 billion active parameters.
  • Trained exclusively on AMD Instinct MI300X clusters with Pensando Pollara networking.
  • Introduces Markovian RSA, a novel test-time compute methodology for unbounded reasoning.
  • Matches or exceeds frontier models like DeepSeek-R1, Gemini-2.5-Pro on several benchmarks.

Highlights and analysis are generated automatically and may contain errors. Check the original source.

More recent stories on this topic · Models