How DeepSeek May Affect AI Agent Tokens
DeepSeek is an AI platform that launched an advanced model (R1), which is said to be able to rival OpenAI’s popular o1 model for much cheaper. This caused immense panic in the tech and financial industry and over $1 trillion was wiped off the US stock market the same day.
DeepSeek stated that they built the model for $6 million, in a few months, and with less-advanced chips. This contradicts Open AI’s estimated $100 million, years of development, and more advanced chips.
But how does this affect AI agents, AI agent tokens, and the broader blockchain market?
What is DeepSeek?
Deepseek (深度求索) is an artificial intelligence company founded in 2023 which has released multiple open-source AI models aimed at AGI development.
The first model released by the company, DeepSeek Coder, was created to assist with coding tasks. Eventually, Deepseek transitioned to more advanced reasoning models, leading to DeepSeek R1, released on January 20, 2025.
Deepseek R1, as the company has claimed, has “performance on par with OpenAI-o1”. The main difference, however, is the R1 model is significantly less expensive than the previously market-dominant OpenAI-o1.
API pricing between the two models
| Deepseek R1 | Open AI-o1 |
| $0.14 / million input tokens (cache hit) | $15 / million input tokens (cache hit) |
| $0.55 / million input tokens (cache miss) | $16.50 / million input tokens (cache miss) |
| $2.19 / million output tokens | $60 – $66 / million output tokens |
The 99% price difference between the two models has caused rapid disruption in the global financial market, causing some of the most valuable companies in the world including Nividia, Microsoft, and Meta to lose significant stock value.
However, in the world of decentralized finance and blockchain, it’s a different ball game, especially with the relatively new AI agents.
What are AI agents?
Described as a “multi-trillion dollar opportunity” by Jensen Huang, CEO of Nvidia, AI agents are independent cognitive computing systems, capable of carrying out autonomous exchanges and financial decisions while utilizing both centralized and decentralized platforms. However, this article will discuss AI agents backed by blockchain.
AI agents are essentially different from typical AI models such as ChatGPT, DeepSeek R1, and Meta AI because while AI models provide recommendations, these agents can make autonomous decisions within the set parameters determined by their programmers, based on available data.
However, it is important to note that AI agents are mostly powered by AI models.
Think of it this way: an AI model is a passenger and you are the driver. You make turns and stops based on the passenger’s recommendation. However, with an AI agent, you are the passenger and the agent is the driver. You tell the driver your destination, and the driver takes you there based on their experience, knowledge, and skill.
AI agents are used in Web 3 for trading, risk management, automated decision-making, and so on.
A notable example of AI agents in Web 3 is Truth Terminal and its unconventional posts on X, including the promotion of a meme coin, Goatseus Maximus ($GOAT). $GOAT went on to achieve a market cap of close to $1 billion.
However, this is not all that AI agents do on the blockchain:
- AIXBT, for instance, is most notable for market analysis
- Zerebro makes art and publishes music on Spotify
- Ripple improves blockchain development on the XRP Ledger
- Theoriq helps simplify web3 development and so on.
Therefore, AI agents on the blockchain have various real-use cases.
What are AI agent tokens?
AI agent tokens are cryptocurrency assets connected to AI agents. While agents perform different autonomous functions, AI agent tokens can be used to pay for these functions, provide economic utility to programmers and other holders, and facilitate transactions in the AI ecosystems. They can also be used to reward contributors to the development of AI in blockchain.
How can DeepSeek affect AI agents and AI agent tokens?
Following the R1 launch by DeepSeek, the cumulative AI agent token market cap declined from $12.9 billion to $10 billion within 24 hours.
More specifically, some popular AI tokens and AI-affiliated tokens such as AIXBT dropped in value by 18%, Freysa AI by 15.9%, Origin Trail by 13.8%, and so on. The top AI agent platforms aI16z and Virtuals Protocol dropped by 24% and 14% respectively. This is according to CoinGecko
This fall was due to widespread uncertainty in the decentralized finance artificial intelligence (DeFAI) market, now that AI agents can be built from a much cheaper alternative.
In addition to being cheap, DeepSeek is also open-source, which means more advanced AI agents can be built for less. This raises the question of whether the AI tokens are as valuable as initially perceived.
To stay on top of the market, the DeFAI sector may have to consider:
- Lowering costs and adjusting pricing
- Adopting DeepSeek’s R1 model as a foundational framework for training AI agents
- Prioritizing AI agent token utility above the hype
- Encouraging contributions and improved integration between AI and the blockchain.
What we think about the future of AI agent tokens
The future of DeFAI seems to ascend, with great potential and room for growth, and as AI continues to improve, more advanced and perhaps cheaper AI models than DeepSeek’s R1 will emerge. However, the sector will have to learn to adapt to prevent the market uncertainty and disruption that came with the release of a cheaper, more advanced AI model.
To do this, AI agent tokens need to provide more real and tangible use cases in the blockchain, improving their economic utility. And with R1’s open-source design, innovation and collaboration on the blockchain cannot be more likely.
Please be advised, that this article or any information on this site is not an investment advice, you shall act at your own risk and, if necessary, receive a professional advice before making any investment decisions
