Nous Research

2026•Nous Research

Hermes Agent v0.12.0 vv0.12.0

Hermes Agent v0.12.0 is Nous Research's open-source autonomous agent that lives on your server, remembers what it learns over time, and integrates with Telegram, Discord, Slack, WhatsApp, Signal, Email, and CLI for multi-platform continuous operation.

Persistent memory and auto-generated skills that improve the longer the agent runs

Natural language cron scheduling for reports, backups, and briefings running unattended

Delegates and parallelizes via isolated subagents with their own terminals and Python RPC scripts

PricingFree (Open Source, MIT License)

Versionv0.12.0

2026•Nous Research

Hermes Agent v0.17.0 v0.17.0

Open Source

AI agent

automation

Hermes Agent v0.17.0 introduces significant enhancements, including integration with iMessage via Photon Spectrum, addition to the Raft agent network, and a more capable desktop application. It also offers background subagents, image editing capabilities, and automation blueprints for scheduling tasks without learning cron syntax.

iMessage integration via Photon Spectrum

Raft agent network addition

Enhanced desktop application with rebindable shortcuts and native notifications

PricingFree

Version0.17.0

2026•Nous Research

TST — Token Superposition Training vv1.0 (Research)

Paid

LLM Training

Efficiency

Nous Research unveiled Token Superposition Training (TST), a technique claiming 2.5x faster LLM training times by optimizing how models process redundant data sequences. TST works by superimposing tokens during training to reduce compute on repetitive patterns, with zero architectural changes required to the final deployed model.

2.5x faster LLM pretraining with no changes to the final model architecture

Token superposition reduces compute on redundant sequence patterns

Compatible with standard transformer architectures and training pipelines

PricingFree (Research Release)

Versionv1.0 (Research)

2026•Nous Research

NousCoder-14B v1.0

Open Source

LLMs

NousCoder-14B is a competitive programming model developed by Nous Research, fine-tuned from Qwen3-14B using reinforcement learning. It achieved a Pass@1 accuracy of 67.87% on LiveCodeBench v6, surpassing the baseline accuracy of 60.79% from Qwen3-14B. The model was trained on 24,000 verifiable coding problems over four days using 48 B200 GPUs.

Competitive programming specialization

Post-trained on Qwen3-14B via reinforcement learning

Achieved 67.87% Pass@1 accuracy on LiveCodeBench v6

PricingFree

Version1.0

2026•Nous Research

Hermes Agent v0.13.0 vv0.13.0

Paid

AI Agents

Open Source

Hermes Agent v0.13.0 — The Tenacity Release — is a self-improving AI agent that finishes what it starts. New features include a durable multi-agent Kanban board with heartbeat monitoring and zombie-agent recovery, a Fast Mode for speed optimization, and Android support for mobile deployment.

Durable multi-agent Kanban board with heartbeat and zombie-agent recovery

New Fast Mode for OpenAI API speed optimization

Android support added for mobile deployment

PricingFree (Open Source)

Versionv0.13.0

2026•Nous Research

Hermes Agent v1.0.0

Hermes Agent is an open-source, autonomous personal assistant designed to integrate seamlessly with various messaging platforms, including Telegram, Discord, Slack, WhatsApp, and command-line interfaces. It operates directly on your server, learning from your interactions and projects to enhance its capabilities over time. By installing Hermes Agent, users gain a persistent assistant that adapts to their needs, offering features like web search, browser automation, and multi-model reasoning. ([nousresearch.com](https://nousresearch.com/hermes-agent/?utm_source=openai))

Integration with multiple messaging platforms

Persistent memory across sessions

Built-in cron scheduler for automations

PricingFree

Version1.0.0