grady

SuperGrok now in Hermes Agent

Run @NousResearch's Hermes Agent fully locally on DGX Spark. 🚀 Our newest playbook shows you how to get set up via @Ollama step by step. 👇

Today we release Lighthouse Attention, a selection-based hierarchical attention for long-context pre-training that delivers a 1.4-1.7× wall-clock speedup at 98K context. It runs the same forward+backward pass ~17× faster than standard attention at 512K context on a single B200, without a custom sparse attention kernel, a straight-through estimator, or an auxiliary loss. During training, queries, keys, and values are pooled symmetrically into a multi-resolution pyramid. We then score every pyramid heads, and a top-k cascade selects a small hierarchical dense sub-sequence, and after a sorting pass that enforces causality, we use standard attention for token mixing. A brief full attention resume at the end converts the checkpoint back into a competent dense-attention model. Validated this using 530M parameter Llama-3 models across 50B tokens, with up to 1M-token benchmarks across 32 B200s under context parallelism. The work on Lighthouse Attention was led by @bloc97_, @SubhoGhosh02, and @theemozilla.

Deepseek V4 Flash is now free via Nous Portal for a limited time thanks to @novita_labs!

Congrats to our team, users, and supporters on the 150,000 GitHub star milestone!

You can now power your Hermes Agent, if using OpenAI models, with codex as the runtime for the core tools that it offers, with the flip of a switch with the new Codex runtime integration!

if you're an artist or creative wondering how to use agents to accelerate your work, or skeptical that they can: please stop what you are doing and look at the 6 examples in this thread

The Hermes Agent Creative Hackathon sponsored by @Kimi_Moonshot has ended! Finalists were selected by Nous and Kimi staff out of 227 submissions on creativity, usefulness and presentation. We were absolutely blown away by the creativity of the things you all built using Hermes. It was tough choosing winners, huge thank you to all who participated! Winners below:

@NousResearch in the lab

Today we release Token Superposition Training (TST), a modification to the standard LLM pretraining loop that produces a 2-3× wall-clock speedup at matched FLOPs without changing the model architecture, optimizer, tokenizer, or training data. During the first third of training, the model reads and predicts contiguous bags of tokens, averaging their embeddings on the input side and predicting the next bag with a modified cross-entropy on the output side. For the remainder of the run, it trains normally on next-token prediction. The inference-time model is identical to one produced by conventional pretraining. Validated at 270M, 600M, and 3B dense scales, and at 10B-A1B MoE. The work on TST was led by @bloc97_, @gigant_theo, and @theemozilla.

Run powerful, self-improving AI agents from your desk. @NousResearch's Hermes Agent brings reliable, self-evolving agentic AI to NVIDIA RTX PCs and DGX Spark. Get started. 👇

🚀Qwen3.6-Plus is on Nous Portal now and FREE for a limited time. Hermes Agent, here we go!! ⚡️ @NousResearch

The future of AI is agent-native. Excited to kick off this journey together with Hermes Agent and the @NousResearch community. Qwen 3.6 Plus is now FREE for a limited time on Nous Portal — give it a try. ⚡

Save the date for May 29 at 6:30pm. Be here in Montreal or be square. You are cordially invited to a cozy apartment MEETUP hosted by myself and @vncntxyz to discuss all things Hermes Agent, everyone's favourite autonomous agent. We'll also be celebrating the end of May and welcome the arrival of another lovely month of June in this city, so bring your good energy and excellent vibes! Food, (maple) water and merch to be served. We know you love Nous merch! Generously sponsored by @NousResearch and Maple3 water. There is a 30 people max capacity, so you MUST sign up below in this specific form and let us know you'll be there to confirm so we can add you to the invite list and do the head count. Tell us what you would like for us to discuss too! cc @gradyacc docs.google.com/forms/d/1XoVEu…

Thanks to everyone who showcased their projects in today's Hermes Agent Jam!

Being shown off in Hermes Jam Session right now 👀👀👀

Step 3.5 Flash from @StepFun_ai is now free again in Nous Portal for the next 15 days!

Computer use with any model Hermes Agent × @trycua