Teknium

Open-source work

Teknium's GitHub profile describes work in AI and data engineering: building open-source AI agents and releasing open language models, datasets, synthetic data pipelines and reinforcement learning environments. At Nous Research, the profile lists contributions to the Hermes model series, from Nous-Hermes-Llama2-13b and the Hermes 2 models on Mixtral-8x7B and Yi-34B to the Hermes 3 and Hermes 4 series, along with Atropos, a framework for reinforcement learning environments with LLMs.¹

Teknium is first-listed author of the Hermes 3 technical report, released in August 2024, and the Hermes 4 technical report, released in August 2025, which describes a family of hybrid reasoning models. On a personal Hugging Face account, Teknium has also released the OpenHermes models, including OpenHermes 2.5 Mistral 7B.²³¹

Products and projects

Teknium names Hermes Agent as the current focus: a terminal-native, extensible coding and personal agent with persistent memory, skills and gateways to messaging platforms, which runs locally and works with major LLM providers. The Hermes Agent repository is published by Nous Research under the MIT license.¹⁴

Work and contributions

• Hermes

  First-listed author of the Hermes 3 and Hermes 4 technical reports; the GitHub profile lists the Hermes 2, 3, and 4 series among Teknium's work at Nous Research.²³¹

• Hermes Agent

  Named on the GitHub profile as Teknium's current focus, described there as an extensible terminal agent with persistent memory that runs locally.¹

• Hermes 4 70B Open weights

  First author of the Hermes 4 Technical Report, which the Hermes 4 70B model card links and asks users to cite.³⁵

• Hermes 4.3 36B Open weights

  First author of the Hermes 4 Technical Report, which the Hermes 4.3 36B model card links and asks users to cite.³⁶

Timeline

1. Aug 15, 2024Released the Hermes 3 Technical Report with Nous Research as first-listed author.²
2. Aug 25, 2025Released the Hermes 4 Technical Report on a family of hybrid reasoning models as first-listed author.³

Where they work

• Nous Research 2 people in the wiki

Related people

• Jeffrey QuesnelleAlso at Nous Research
• Vasi PhilominLanguage models · Agents
• Jason WeiLanguage models · Agents
• Yoav ShohamLanguage models · Agents
• Thomas ScialomLanguage models · Agents
• Binyuan HuiAgents · Language models
• Qingyun WuAgents · Language models
• May HabibLanguage models · Agents
• Harrison ChaseAgents · Language models
• Hanlin TangLanguage models · Agents