Anthropic has launched its next generation of Claude models, and one in particular stands out as a developer’s dream.
Claude Opus 4, unveiled alongside Claude Sonnet 4, sets “new standards for coding”, according to the startup, and marks its most powerful model launch yet.
Opus 4 is designed specifically for software developers and engineers and “excels at coding and complex problem-solving” tasks.
The company claims the model achieved a 72.5% score on SWE-bench, which is used to benchmark software engineering tasks. Notably, this means the model dramatically outperforms OpenAI’s GPT-4.1, which scored 54.6% on the same testing.
When OpenAI announced the launch of GPT-4.1, it targeted the model at software developers - an area which the company has focused on in recent months.
Admittedly, this model did mark a sizable improvement compared to GPT-4o, which scored 21.4% when it launched. Regardless, Anthropic appears to have once again blown it out the water.
“These models advance our customers' AI strategies across the board: Opus 4 pushes boundaries in coding, research, writing, and scientific discovery, while Sonnet 4 brings frontier performance to everyday use cases as an instant upgrade from Sonnet 3.7,” the firm said in a blog post announcing the launch.
Elsewhere, Claude Opus 4 significantly outperforms previous models on memory capabilities, according to Anthropic. The firm said that when developers building applications provide Claude access to local files, the model “becomes skilled” at creating and maintaining memory files.
This allows the model to store key information more efficiently, thereby improving long-term task awareness, performance, and coherence.
Claude Opus 4 has serious stamina
According to Anthropic, the Opus 4 model boasts a combination of performance and stamina, so to speak.
During testing at Rakuten, the model contended with a “demanding” open source refactoring exercise while running independently for seven hours. Notably, Anthropic said it did so with “sustained performance”.
This marks a step change in performance and longevity, and suggests developers harnessing Opus 4 can do so across the length of their working day, rather than in sporadic bursts of activity.
“Cognition notes Opus 4 excels at solving complex challenges that other models can't, successfully handling critical actions that previous models have missed,” Anthropic said in its blog post.
How to access the new models
Anthropic described Opus 4 and Sonnet 4 as “hybrid models” which offer two distinct modes. These include “near-instant responses and extended thinking for deeper reasoning”.
The Pro, Max, Team, and Enterprise Claude plans will include both models and extended thinking capabilities. Notably, Sonnet 4 will also be available for free users.
Pricing for Opus 4 starts at $15/$75 per million tokens (input/output) and Sonnet 4 at $3/$15.
Both will be made available via the Anthropic API, or through Amazon Bedrock and Google Cloud’s Vertex AI service.
MORE FROM ITPRO
- Anthropic ramps up European expansion with fresh hiring spree
- There's no single route to AI adoption, and Anthropic is charting all available paths
- Databricks and Anthropic are teaming up on agentic AI development
-
Otter.ai wants to bring agents to all third party systems – with transcription just the startNews The AI transcription company is targeting intelligent scheduling and interoperability with project management systems, based on securely-stored transcription data
-
OpenAI signs another chip deal, this time with AMDnews AMD deal is worth billions, and follows a similar partnership with Nvidia last month
-
AI coding really isn't living up to expectations – "the savings have been unremarkable" but not for the reason you might thinkNews Companies are focusing too heavily on simple AI coding tasks, and not overhauling wider business processes
-
UK government programmers trialed AI coding assistants from Microsoft, GitHub, and Google – here's what they foundNews Developers participating in a trial of AI coding tools from Google, Microsoft, and GitHub reported big time savings, with 58% saying they now couldn't work without them.
-
Microsoft touts new Copilot features in Excel, but says you shouldn’t use them if you want accurate resultsNews Microsoft has warned against using new AI features in Excel for “tasks with legal, regulatory, or compliance implications” – so when can you use it?
-
Senior developers are all in on vibe coding, but junior staff lack the experience to spot critical flawsNews Experienced developers are far more confident in using AI-generated code
-
Hexaware partners with Replit to take secure 'vibe coding' to the enterpriseNews The new collaboration enables business teams to create secure, production-grade applications without the need for traditional coding skills
-
Microsoft says AI is finally having a 'meaningful impact' on developer productivity – and 80% 'would be sad if they could no longer use it'News Researchers at Microsoft wanted to demystify how AI is being used by software developers – their findings show the benefits are finally becoming clear.
-
AWS CEO Matt Garman just said what everyone is thinking about AI replacing software developersNews Junior developers aren’t going anywhere, according to AWS CEO Matt Garman
-
Google's new Jules coding agent is free to use for anyone – and it just got a big update to prevent bad code outputNews Jules came out of beta and launched publicly earlier this month, but it's already had a big update aimed at improving code quality and safety.
