Products
Developers
About

Insights & Information

Find what you need to accelerate your AI journey

Blog

Why SambaNova's SN40L Chip is The Best for Inference

Comparing the end-user inference performance of SambaNova's technology against that of Groq and Cerebras.

Blog

SubgoalXL: Pushing the Boundaries of LLM in Formal Theorem Proving

SubgoalXL represents a significant step forward in the field of AI-powered theorem proving.

Blog

SambaNova Holds Speed Record on Llama 3.1 405B - 4X faster than the rest

Today, we’ve set a world performance record of 114 tokens per second on Llama 3.1 405B, independently verified by...

Blog

Three Predictions for the Upcoming Llama 3 405B Announcement

Three predictions on how Llama 3 405B could reshape the landscape for developers engaged in AI and machine learning.

Blog

Typhoon model adds Thai language to Samba-1

With the inclusion of Typhoon Thai LLM, Samba-1 is now able to deliver generative AI capabilities in the Thai...

Blog

Does reduced precision hurt? A bit about losing bits.

Recent work highlighted how quantization for recent LLaMa 3 models can lead to non-negligible decay in model...

Blog

SambaNova CEO explains why only one AI company wants a monopoly

Rodrigo Liang and veteran tech journalist Don Clark of The New York Times discussed how a full stack approach to AI...

Blog

Transform Your Data Privacy with SambaNova Systems

Samba-1 provides role-based access controls to maintain data governance policies, ensuring only those with proper...

Blog

SambaNova has broken the 1000 t/s barrier: why it's a big deal for enterprise AI

SambaNova is the clear winner of the latest large language model (LLM) benchmark by Artificial Analysis. Topping the...

Blog

Model Ownership

As enterprises incorporate generative AI into their business, retaining model ownership is one of the most important...

Blog

Introducing Fugaku-LLM in Composition of Experts

The Fugaku-LLM, a Japanese LLM, is being introduced into the Samba-1 CoE architecture to run optimally on the...

Blog

Sovereign AI

A Sovereign AI solution is fully contained with an entity, such as a company or country, and meets objectives while...

Blog

NAIRR; Govt-funded AI Research Resources

NAIRR pilot, in partnership with SambaNova, provides generative AI platforms for groundbreaking academic research....

Blog

Tokens Per Second is Not All You Need

In this post, we explore why tokens per second doesn't paint the full picture of enterprise LLM inference...

Blog

The Next Generation of Large Models

Generative AI can streamline processes across the entire organization, reduce costs, increase productivity, improve...

Blog

Enterprise-grade AI

See what it means to be enterprise-grade, and how SambaNova Suite, the first full-stack platform, purpose-built for...

Blog

Samba-CoE v0.3: The Power of Routing ML Models at Scale

Samba-CoE-v0.3, our latest Composition of Experts, surpasses DBRX Instruct 132B and Grok-1 314B on the OpenLLM...

Blog

Responsible AI

SambaNova is committed to providing customers with responsible, generative AI that is safe, secure, and transparent,...

Blog

SambaLingo hits 15,000+ downloads, now integrated with Samba-CoE-v0.2

SambaLingo has been downloaded over 15,000 times and has achieved remarkable performance of 280 tokens/s inference...

Blog

SambaNova Delivers Accurate Models At Blazing Speed

Samba-CoE v0.2 is climbing on the AlpacaEval leaderboard, outperforming all of the latest open-source models.

Blog

Using Mixed Precision on RDUs

SambaFlow 1.18 introduces support for mixed precision on RDUs, streamlining the experience for model developers and...

Blog

Sambaverse: Discover, Compare, Evaluate

Sambaverse is a unique environment where developers can freely test out hundreds of different models and directly...

Blog

Benchmarking Samba-1

Benchmarking Samba-1 with the EGAI benchmark - a comprehensive collection of widely adapted benchmarks sourced from...

Blog

Samba-CoE v0.1 - Unlocking the power of routing to build a Composition of Experts

We're thrilled to unveil Samba-CoE-v0.1, a scaled down version of Samba-1, our latest breakthrough model that...