The company was formed in 2021 by brother and sister Dario and Daniela Amodei, after they left OpenAI.
ADVERTISEMENTArtificial intelligence (AI) company Anthropic has launched its fastest and most powerful large language model (LLM) as the race to dominate the technology heats up.
The company, which develops general AI systems, was founded by former members of OpenAI and is backed by Amazon, Google, and Salesforce among other companies. LLMs are the AI program that recognises and generates text, among other tasks
On Monday it released its so-called Claude 3 model family, which includes three state-of-the-art models in ascending order of capability: Claude 3 Haiku, Claude 3 Sonnet, and Claude 3 Opus.
Anthropic claimed the most capable of the models Opus, outperformed OpenAI’s GPT-4 and Google’s Gemini Ultra on tests such as reasoning, basic maths, and undergraduate and graduate-level knowledge.
Opus and Sonnet are now available to use in claude.ai and the Claude API which is now generally available in 159 countries.
The company said the more compact models, Sonnet and Haiku, will be released soon.
Opus can summarise approximately 200,000 tokens, while ChatGPT can only summarise about 3,000. A token is a unit of measurement that calculates the amount of text data that is processed by your chatbot
The new models also allow images and documents to be uploaded for the first time.
The company was formed in 2021 by brother and sister Dario and Daniela Amodei, after they left OpenAI. In an interview with Fortune, Dario said he left the company due to the lack of attention OpenAI paid to safety.
Anthropic said it had several dedicated teams to track and mitigate risks such as misinformation, biological misuse, and election interference. It also said Claude 3 shows fewer biases than previous models.
The company was named Anthropic to show that humans (anthro) are at the centre of the AI firm. Anthropic is also listed as a public-benefit corporation, meaning they have to report how the company has promoted its benefit to the public.
Anthropic said companies like Airtable and Asana helped A/B test the models.
“We’re enthusiastic about this release after seeing very promising improvements in latency: our initial tests are showing a 42 per cent improvement in time-to-first-token, and over a 200 per cent improvement in tokens/second,” said Eric Pelz, Head of Technology, AI at software company Asana.
“We’re also observing an improvement in following the nuances of our prompts, all while maintaining output quality”.
