Anthropic has launched Claude Opus 4 and Claude Sonnet 4, its newest technology of hybrid-reasoning AI fashions optimized for coding duties and fixing advanced issues.
Claude Opus 4 is Anthropic’s strongest AI mannequin so far, based on the corporate’s announcement, and able to working constantly on long-running duties for “a number of hours.” In buyer checks, Anthropic mentioned that Opus 4 carried out autonomously for seven hours, considerably expanding the possibilities for AI agents. The corporate additionally described its new flagship because the “finest coding mannequin on this planet,” with Anthropic’s benchmarks exhibiting that Opus 4 outperformed Google’s Gemini 2.5 Pro, OpenAI’s o3 reasoning, and GPT-4.1 models in coding duties and utilizing “instruments” like net search.
Claude Sonnet 4 is a extra reasonably priced and efficiency-focused mannequin that’s higher suited to basic duties, which supersedes the 3.7 Sonnet model released in February. Anthropic says Sonnet 4 delivers “superior coding and reasoning” whereas offering extra exact responses. The corporate provides that each fashions are 65 % much less more likely to take shortcuts and loopholes to finish duties in comparison with 3.7 Sonnet and so they’re higher at storing key data for long-term duties when builders present Claude with native file entry.
A brand new function launched for each Claude 4 fashions is “considering summaries,” which condenses the chatbots’ reasoning course of into simply comprehensible insights. An “prolonged considering” function can also be launching in beta that enables customers to change the fashions between modes for reasoning or utilizing instruments to enhance the efficiency and accuracy of responses.
Claude Opus 4 and Sonnet 4 can be found on the Anthropic API, Amazon Bedrock, and Google Cloud’s Vertex AI platform, and each fashions are included in paid Claude plans alongside the prolonged considering beta function. Free customers can solely entry Claude Sonnet 4 for now.
Along with the brand new fashions, Anthropic’s Claude Code agentic command-line device is now typically out there following its restricted preview in February. Anthropic additionally says it’s shifting to supply “extra frequent mannequin updates,” as the corporate tries to maintain up with competitors from OpenAI, Google, and Meta.