At Essential AI, we're building an open platform to democratize frontier AI capabilities and accelerate breakthroughs globally through collaborative science.
Today, we’re excited to introduce Rnj-1, @essential_ai's first open model; a world-class 8B base + instruct pair, built with scientific rigor, intentional design, and a belief that the advancement and equitable distribution of AI depend on building in the open.
We bring American open-source at par with the best in the world.
Farbod Tavakkoli from @ATT and @DiamosGreg from @tensorwave demonstrated world-class efficiency at scale, achieving exceptional utilization across 256 @AMD MI325X GPUs on TensorWave using ScalarLM.
This is a victory for American open weights models.
[1/5]
We thank the community for their feedback on Rnj-1. We’d like to announce a few updates to Rnj-1-instruct based on what we heard:
- Resolving premature truncation of generations and improved instruction following.
- Instructions for 128k context length extrapolation.
- Updated evals, baselines, and model generations for reproducibility.
Details follow 🧵
[4/5]
Our tables have been updated with the Olmo-3-7B-Instruct baseline, and we have added post-processing improvements to our STEM evals, boosting performance. For reproducibility, we release our eval prompts and generations.
Proud to carry the spirit of Ramanujan with rnj-1: honoring the great minds who showed how curiosity reshapes the world. The lineage of deep thinking is part of our mission. Essential proud.
Behind the scenes at Essential: long hours, tough problems, and a team driven by an unsatiable desire to build.
What keeps us going isn’t noise - it’s the scientific discipline. The work demands focus, rigor, and a willingness to push through uncertainty.
Proud to work with this special team @essential_ai@ashVaswani
@essential_ai Highlights:
👉 Dominant on SWE-bench, BFCL, and Enamel
👉 Strong tool use and function calling
👉 32K context window
👉 Apache 2.0 licensed, fully open weights
👉 Trained on 8.4T tokens with Muon optimizer
Introducing Rnj-1 Instruct from @essential_ai, an open-source 8B model engineered for agentic coding and STEM tasks. AI natives can now use Rnj-1 Instruct on Together AI and benefit from reliable inference for production-scale software engineering and scientific workflows.
We are beyond thrilled to share our first flagship models, Rnj-1 base and instruct 8B parameter models. Rnj-1 is the culmination of 10 months of hard work by a phenomenal team, dedicated to advancing American SOTA OSS AI.
Lots of wins with Rnj-1.
1. SWE bench performance close to GPT 4o.
2. Tool use outperforming all comparable open source models.
3. Mathematical reasoning (AIME’25) nearly at par with GPT OSS MoE 20B.
….