Wen Chen Gen.
In the past two days, another big thing has happened in the field of AI. That's when OpenAI's rival company Anthropic released their new model, the Claude 3, and claimed to surpass GPT-4 across the board. Overnight, the news swept the Internet.
After all, ChatGPT is a product that really ignites the large-scale model market, especially in the past year, the news about ChatGPT and GPT-4 has not stopped, either getting high scores in what tests, or being applied in which industry. Of course, in the past year, there have also been many tech giants that have launched their own competing products, wanting to surpass OpenAI and get a piece of this new AI wave, but none of them have been very successful.
Well, this time the claim is fully superior to GPT-4claude 3What's going on? Is it really beyond GPT-4, or is it yet another misnomerous propaganda?
Unlike in the past, this time, GPT-4 may really be surpassed. A lot of people may not know about Anthropic, the company that launched Claude 3. In fact, OpenAI andanthropicThe founding of the company is very closely related. In 2020, due to the differences in values among OpenAI's internal executives, some people felt that they should support commercialization and accelerate the development of AI, but some people believed that it was necessary to develop safer AI, including Dario Amode and Daniela Amodei. So, the two siblings left OpenAI and founded Anthropic.
Today, Anthropic is just over three years old. However, although it is a start-up, its strength is not weak at all. If OpenAI is the industry leader,StartupsanthropicthenIt's the second one who does not give upEven beenThe strongest competitor called OpenAI
On November 30, 2022, ChatGPT was released, and less than four months later, in March 2023, Anthropic launched a ChatGPT-like ChatGPT-like product, which was also the fastest ChatGPT-like product launched in the world at that time, and even Google's speed was not as fast as Anthropic.
The Claude 3 is actually the third generation of the Claude series. The new Claude 3 series consists of three models, in order of ability: Claude 3 Haiku, Claude 3 Sonnet, and Claude 3 Opus.
The Claude 3 OPUS is the smartest model, outscoring GPT-4 and Gemini 1 in multiple benchmarks0 ultra，It has set a new industry benchmark in mathematics, programming, multilingual understanding, vision and other dimensions. Anthropic said Claude 3 Opus has human-undergraduate-level knowledge.
If we play the role of an economist in Claude 3, it can also give a very professional analysis in front of open-ended questions. For example, send Claude 3 a chart of the GDP of the United States over the past two decades, and let it show the general direction of the U.S. economy in the next few years. In just a few seconds, it not only produced results, but also produced dozens of directions.
claudeFor the first time, the series also has the ability to be multimodalThe Claude 3 has sophisticated visual features comparable to other head models. They can handle data in a variety of visual formats, including charts, graphs, and technical charts. Users can now upload **, charts, documents, and other types of unstructured data for Claude3 to analyze and answer.
What's more, Anthropic believes that Claude 3 can be trusted. Anthropic's multiple dedicated teams are responsible for tracking and mitigating risks, which is what Anthropic's future of AI has been advocating for and wants to build.
At present, Anthropic's valuation has reached $30 billion, making it the second largest model company after OpenAI. And the competition for large models may continue, and now, the winner is undecided.