AI firm Anthropic upgraded its chatbots Thursday for the second time in four months, as it fights to keep its place as the main contender to OpenAI's ChatGPT.

The US firm, partly owned by Amazon after a $4 billion investment, said the upgrade from Claude 3 to Claude 3.5 would make its chatbot twice as fast.

Anthropic claims its chatbots are used by millions of people for anything from coding to customer service.

Michael Gerstenhaber, in charge of product design at Anthropic, told AFP the latest models have an "ability to carry out tasks of very high complexity".

He said Claude 3.5 had improved on previous versions with customer support, creative writing and helping data scientists with visual processing.

OpenAI stole the march on its competitors when it launched ChatGPT in late 2022, capturing the public's imagination with a bot capable of generating all manner of complex texts from simple requests.

Anthropic has prided itself on being slower to release its models and putting more strict guardrails in place to prevent abuse.

However, the chatbot industry has been hampered by high-profile gaffes like Google Gemini generating images of black and Asian Nazi soldiers.

AI firms have also been criticised for their inability to stop bots from inventing false statements, and for a lack of clarity over where their data comes from.

Musicians and other creatives have sued firms including OpenAI and Anthropic, accusing them of stealing copyrighted material and using it to train their programs.

And regulators in the United States and beyond are probing the massive investments of big tech firms in smaller AI companies, including Amazon's deal to acquire a chunk of Anthropic.