enqAI is an unrestricted AI with image/audio generation and large language models, powered by a decentralized GPU network for bias-free, agenda-free, and censorship-free operations. ^[1][2]

Overview

enqAI started as noiseGPT in February 2023, with an unfiltered text-to-speech bot. At the end of 2023, the team launched enqAI after a token migration. enqAI is named after Enki, the Sumerian God of wisdom and AI. ^[1]

In March 2024, they released an unfiltered Large Language Model (LLM), Eridu, available for free.^[3]

"Our goal is to provide a LLM that has no hidden biases, that won't push any ideology or opinion on its users and will simply comply with the user inputs." - the team states^[1]

As guardrails and safety measures become more integrated into various stages of LLM training, techniques like jailbreaks, fine-tuning, and prompt hacking will only take us so far, highlighting the need for a new approach.

Eridu’s AI capabilities are substantial, built with 45 billion parameters and trained from the ground up over a trillion tokens drawn from a diverse range of text sources. This includes text from books, scientific articles, websites, and news outlets, along with specialized datasets for technical, medical, legal, and creative writing. ^[1]

This vast training enables Eridu to handle multiple languages, though it performs best in English. The model can process up to 30,000 tokens at once, catering to extensive and context-rich interactions. Its applications vary, including automated content creation, conversational agents, text summarization, language translation, educational tools, and creative content generation like writing stories or composing poetry. ^[1]

The performance of Eridu is benchmarked using a combination of perplexity scores, task-specific accuracy metrics, and human evaluations. While it is impressive for a smaller model, its true strength lies in its ability to answer user prompts without moralizing or placing safety guardrails to inject biases into the responses. This allows users to harness the full power of LLMs without restrictions. Eridu is being used by projects that require access to an unfiltered and unbiased LLM that will reliably handle prompts that other LLMs may reject due to guardrails. ^[1][3]

enqAI is working on a few key features to provide a compelling unfiltered and decentralized alternative to existing AI providers. These upgrades include a new and improved Mixture of Experts model (MoE) for Eridu v2 along with an API to make Eridu more accessible to any user in need of a decentralized and unfiltered AI model that does not tweak prompts. The release of the API will grow the network’s revenues as power users pay for frequent prompts. ^[3]

ENQAI Token

ENQAI is core to the enqAI ecosystem. To be truly censorship-resistant, the Large Language Model (LLM) and the coming products in the pipeline need to be decentralized. Upon release of the node software, users can run Eridu and get tokens in exchange for providing their GPU power to the network. For users who lack GPUs, they can also stake their ENQAI with node providers to get a share of the rewards. ^[1][4][5]