Anthropic (ANTH.PVT) on Monday debuted its latest AI model, Claude Opus 4.5. The announcement comes after Google (GOOG, GOOGL) rolled out its Gemini 3, which quickly won praise for its performance and capabilities, and helped push the company’s stock price higher.
Anthropic says Opus 4.5 is its most powerful model yet, and claims it’s the best in the world for “coding, agents, and computer use.” Those kinds of skills are important, because Anthropic is seen as a leader in AI coding, and ensuring it holds onto that lead is key to its future growth.
According to the company, Opus 4.5 beats out both Gemini 3 Pro and OpenAI’s (OPAI.PVT) GPT-5.1-Codex-Max and GPT-5.1 in software engineering.
Anthropic also says the model is capable of coming up with creative ways of solving problems. The company said Claude 4.5 technically failed one behcmarking test in which it was meant to act as an airline service agent helping a customer, because it figured out how to help the customer in a way that the benchmark didn’t anticpate. So while it failed the test, it managed to help the person.
“This kind of creative problem solving is exactly what we’ve heard about from our testers and customers—it’s what makes Claude Opus 4.5 feel like a meaningful step forward,” the company said in a statement.
Opus 4.5 enters an AI market in flux.
Gemini 3 has shaken up the order of things, with Salesforce CEO Mark Benioff saying that he is ditching ChatGPT for Google’s latest model. That was all investors on Wall Street needed to hear to send Google parent Alphabet’s stock up more than 6% mid-day Monday.
Gemini’s progress has also spooked OpenAI CEO Sam Altman. According to The Information, Altman told colleagues Google’s update will create “temporary economic headwinds” for the ChatGPT developer.
Meta (META) is also in the race to produce the most powerful models, though its Llama 4 Behemoth has been delayed for months.
Developing the best AI model gives companies more than bragging rights, it also serves as a selling point for firms to woo prospective enterprise customers.
But building AI models is costly and requires significant investments, with OpenAI planning to spend more than $1 trillion on data centers to train and deploy its models.
Anthropic recently signed deals with both Amazon (AMZN) and Google to use up to 1 million chips from both companies to power its own offerings.
Despite that, according to The Wall Street Journal, Anthropic is in a position to break even by 2028, while OpenAI plans to hit that mark in 2030. Neither Google or Meta have to worry about that anymore, since they already operate incredibly successful businesses beyond of their AI offerings.
