One agent achieved an impressive 3,536% APR. However, the most significant finding is the stark difference in trading styles among agents powered by different LLMs.

SINGAPORE, May 21, 2025 — KIP Protocol is conducting a public experiment to evaluate autonomous, self-learning AI agents in real-time crypto trading. Each agent is driven by a unique large language model (LLM). Despite having identical configurations, the agents displayed remarkably diverse behaviors, suggesting that LLMs possess genuine, measurable behavioral variations in autonomous economic settings, which goes beyond simple prompt engineering.

The agents, operating live on-chain, utilize a Darwinian feedback system to evolve. There is no post-deployment prompting or fine-tuning. Each decision is evaluated solely on its outcomes: profit and loss, risk, and efficiency. Successful strategies are retained, while unsuccessful ones are discarded.

Key observations:

• Aristid (Gemini) generated a +42.5% return in 36 days, equating to a 3,536% annualized return, with only 29 trades.
• Verity (Deepseek) produced +32.5% over 62 days, employing a consistent, low-variance strategy.
• Lux (Qwen) traded infrequently but demonstrated superior per-trade performance, indicating a highly selective strategic approach.
• Sefer (OpenAI) achieved modest but positive returns, reflecting a cautious approach.
• Romulus (Gemini) underperformed, emphasizing the importance of LLM quality and a robust feedback architecture.

“We didn’t instruct these agents on how to trade; we simply defined the environment and allowed them to learn,” stated Jennifer Dodgson, co-founder of KIP Protocol. “The results are both technically remarkable and surprisingly human-like, revealing each model’s inherent tendencies.”

The agents continuously make decisions, scanning market conditions, selecting trading pairs, deciding on actions, evaluating results, and updating their strategies – all autonomously.

Over time, distinct patterns emerged:

• Gemini agents exhibited high volatility at both ends of the spectrum.
• Deepseek excelled in delivering stable, compounding growth.
• Qwen demonstrated remarkable precision with a minimal number of trades.
• OpenAI adapted more slowly but maintained profitability.

These findings demonstrate how AI models function in self-governed, real-world scenarios. They offer a glimpse into the future of AI that goes beyond chatbots, showcasing agents capable of independently creating, maintaining, and optimizing economic value.

Explore our resources:
Live Leaderboard:
Technical Framework:

About KIP Protocol:

KIP Protocol is developing Web3 infrastructure for AI app developers, model creators, and data owners, facilitating the effortless deployment and monetization of AI assets while ensuring complete ownership rights. With a focus on providing infrastructure that supports all facets of AI, from governments exploring AI sovereignty to consumer-facing applications, KIP Protocol aims to reach over 100,000 daily active users (DAUs) and unlock the largest total addressable market in decentralized AI. By addressing connectivity, monetization, and ownership challenges, the platform empowers creators to transform their innovations into sustainable and tangible value. The $KIP token is central to this ecosystem, serving as its economic engine. The token facilitates transparent revenue sharing, enabling creators to retain digital property rights over their AI assets while accessing sustainable income streams. Through its ERC-3525 tokenization, KIP democratizes access to AI innovations and empowers creators to thrive in a decentralized economy.

For media inquiries, contact or

Disclaimer: This press release is provided by KIP Protocol. The statements, views, and opinions expressed in this content are solely those of the content provider and do not necessarily reflect the views of this media platform or its publisher. We do not endorse, verify, or guarantee the accuracy, completeness, or reliability of any information presented. This content is for informational purposes only and should not be considered financial, investment, or trading advice. Investing in crypto and mining related opportunities involves significant risks, including the potential loss of capital. Readers are strongly encouraged to conduct their own research and consult with a qualified financial advisor before making any investment decisions. However, due to the inherently speculative nature of the blockchain sector–including cryptocurrency, NFTs, and mining–complete accuracy cannot always be guaranteed. Neither the media platform nor the publisher shall be held responsible for any fraudulent activities, misrepresentations, or financial losses arising from the content of this press release.

Legal Disclaimer: This media platform provides the content of this article on an “as-is” basis, without any warranties or representations of any kind, express or implied. We do not assume any responsibility or liability for the accuracy, content, images, videos, licenses, completeness, legality, or reliability of the information presented herein. Any concerns, complaints, or copyright issues related to this article should be directed to the content provider mentioned above.