Original｜Odaily Planet Daily (@OdailyChina)
Following Xiaohongshu's rise to the top of the US AppStore due to the TikTok ban, the AI field has once again seen a "Chinese dark horse" reach the top of the AppStore.
The difference this time is that the value of this achievement far exceeds that of Xiaohongshu—according to media reports, the DeepSeek app has topped the free app download rankings in both the Chinese and US App Stores, surpassing the "benchmark product in the AI field," ChatGPT, in the US download rankings. Meanwhile, the concept tokens related to the Deepseek R1 model have gained popularity due to market attention, with several tokens' market values exceeding the $10 million threshold.
Odaily Planet Daily will provide a brief introduction to the Deepseek product and related concept tokens for readers' reference.
Deepseek: A Direct Challenge to OpenAI, R1 Model More User-Friendly and Affordable than ChatGPT o1
The origin of all this hype lies in an article released by Deepseek's official account seven days ago.
On January 20, an article titled “DeepSeek-R1 Released, Performance Benchmarking OpenAI o1 Official Version” emerged, directly challenging AI giant OpenAI's o1 model, providing a detailed introduction to Deepseek's R1 model, which mentioned:
DeepSeek-R1 extensively uses reinforcement learning techniques in the post-training phase, significantly enhancing the model's inference capabilities with very few labeled data. In tasks such as mathematics, coding, and natural language reasoning, its performance rivals that of OpenAI's o1 official version.
We have made all the training techniques of DeepSeek-R1 public to promote full communication and innovative collaboration within the tech community. Paper link: https://github.com/deepseek-ai/DeepSeek-R1/blob/main/DeepSeek_R1.pdf
The pricing for DeepSeek-R1 API services is 1 yuan per million input tokens (cache hit) / 4 yuan (cache miss), and 16 yuan per million output tokens.
In simple terms, the R1 model launched by Deepseek not only matches the performance of ChatGPT o1 but also directly open-sources the related training techniques, while its API pricing is shockingly low.
While many were still skeptical about whether Deepseek was "bluffing," a series of validation results from industry insiders confirmed: Deepseek is telling the truth! The sharp product that can cover both B-end and C-end needs has left countless AI companies and teams in the US exclaiming, "The wolf is here!"
A team led by Jiayi Pan, a PhD student at the University of California, Berkeley, successfully replicated the key technology "Aha Moment" in DeepSeek R1-Zero at an extremely low cost. The research shows that even smaller language models can develop strong self-validation and search capabilities through reinforcement learning (RL). Notably, the cost of training the model was less than $30. The aforementioned "Aha Moment" is a key technology in DeepSeek R1-Zero.
According to the technical report on DeepSeek-R1, it indicates that through the RL framework, AI may spontaneously form human-like reasoning abilities, even surpassing the limitations of preset rules. This also holds promise for developing more autonomous and adaptive AI models, such as dynamically adjusting strategies in complex decision-making (medical diagnosis, algorithm design).
Previously, a Meta employee posted anonymously on the workplace community TeamBlind in the US, stating that due to the excellent performance and extremely low training costs of DeepSeek's models, Meta engineers are frantically analyzing DeepSeek and trying to replicate some of its content.
Additionally, AI blogger scaling01 conducted a comparative analysis, stating that DeepSeek not only surpassed in performance but also that the training cost of DeepSeek-V3, with a parameter count of 671B, was only $5.58 million. In contrast, the computational cost of Meta's Llama3 series models is enough to train DeepSeek-V3 at least 15 times.
—— Quoted from APPSO related article
It is worth mentioning that Deepseek is backed by the well-known domestic quantitative fund, Huanfang, which previously accumulated a large number of high-performance graphics cards due to quantitative trading needs; and because it is a financial company, it has not been included in the US technology company sanctions list. Ultimately, with extremely strong determination and a long-term technical route, it achieved the current result of "achieving better output effects with a smaller amount of data."
Of course, the meme coins, which have always been a news hotspot, are still leading the market.
Deepseek Related Concept Tokens: Market Hot Coins Intertwined with Technology and Meme
Here are some Deepseek-related concept tokens currently receiving attention in the market.
seek: Same-name AI concept token, market value reported at $22.7 million
As early as December last year, when the Deepseek V3 version was released, many noticed its speculative value. Seek is the earliest "meme coin" that "hopped on" the Deepseek concept, having been created 20 days ago.
According to GMGN data, its current price is reported at $0.022, with a market value of approximately $20.6 million, previously surpassing $33 million, and a 24-hour trading volume of about $54.6 million. It is worth mentioning that Deepseek's official X platform account previouslystated that Deepseek has not released any tokens.
Seek token information
POLAR: AI Copilot built on Deepseek, market value reported at $11 million
Polar's main concept is "the first AI copilot on the Solana ecosystem, built on the Deepseek framework." According to its latest tweet, it has achieved a series of milestone achievements just three days after launch and is about to open the Polar API.
According to GMGN data, its current price is reported at $0.011, with a market value of around $11 million, previously surpassing $22 million, and a 24-hour trading volume of about $36.4 million.
POLAR token information
AAA: Token released by a former Amazon programmer, market value reported at $6.1 million
AAA is a token related to AI Agent released by a former Amazon programmer named nim. According to its tweet information, it uses the Deepseek model and has a cost reduction of 98% compared to the ChatGPT o1 model.
According to GMGN data, its current price is reported at $0.0061, with a market value of around $6.1 million, previously surpassing $7.5 million, and a 24-hour trading volume of about $4.5 million.
AAA token information
Latest news indicates that ai16z founder Shaw stated: "By the way, Eliza completed its support for DeepSeek two weeks ago." It must be said that ai16z is still at the forefront of tracking the AI model race. We will continue to pay attention to more AI concept-related tokens.
Additionally, Odaily kindly reminds: The market for meme coins is highly volatile, and the above content does not constitute investment advice. Please participate cautiously and do your own research (DYOR).
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
