Whenever it seems like artificial intelligence (AI) technology providers have nothing left to launch, they surprise us with something new, and the latest is Anthropic with its next-generation Claude 4 models – Claude Opus 4 and Claude Sonnet 4.
As it happens, AI and large language model (LLM) developer Anthropic has recently released its latest AI models, which it said were “setting new standards for coding, advanced reasoning, and AI agents,” according to the company’s announcement shared on May 22.
Indeed, it unveiled Opus 4, calling it the “world’s best coding model, with sustained performance on complex, long-running tasks and agent workflows” and Sonnet 4 – a “significant upgrade to Claude Sonnet 3.7, delivering superior coding and reasoning while responding more precisely to your instructions.”
Claude Opus 4
Specifically, the most recent developments bring about a set of new features and capabilities, such as Claude Opus 4’s improved memory capabilities in creating and maintaining ‘memory files’ with key information, resulting in “better long-term task awareness, coherence, and performance on agent tasks.”
Among these tasks, for instance, is the creation of a ‘Navigation Guide’ while playing Pokémon, which involves recording key information from local files (when given access) to help improve its gameplay.
Claude Sonnet 4
Meanwhile, Claude Sonnet 4 “balances performance and efficiency for internal and external use cases, with enhanced steerability for greater control over implementations,” and an “optimal mix of capability and practicality,” the company said.
It also excels at autonomous multi-feature app development, following complex instructions, clearer reasoning, problem-solving, codebase navigation (reducing navigation errors from 20% to near zero), output aesthetics, and success rates in everyday use cases as an instant upgrade from Sonnet 3.7.
Other improvements
Alongside the release of Claude 4, Anthropic also introduced thinking summaries, extended thinking with tool use in beta, allowing its AI to alternate between reasoning and tool use to improve responses, and new model capabilities – including the use of tools in parallel, following instructions more precisely, and improved memory.
Then there are the four new API capabilities – the code execution tool, MCP connector, Files API, and the ability to cache prompts for up to one hour, as well as the expansion of how developers can collaborate with Claude, with code supporting background tasks via GitHub Actions and native integrations with VS Code and JetBrains, displaying edits directly in the files.
ReadNext
