Claudes Newest Models are Solid
Claude released its latest new models last week - Claude Opus 4 and Claude Sonnet 4. Anthropic, Claude’s makers, introduced them like so:
Today, we’re introducing the next generation of Claude models: Claude Opus 4 and Claude Sonnet 4, setting new standards for coding, advanced reasoning, and AI agents.
Claude Opus 4 is the world’s best coding model, with sustained performance on complex, long-running tasks and agent workflows. Claude Sonnet 4 is a significant upgrade to Claude Sonnet 3.7, delivering superior coding and reasoning while responding more precisely to your instructions.
It’s no surprise to see Claude Opus 4 touted as the world’s best coding model, as Claude has been know to be a top tier option for coding for a long while now. The mention of “sustained performance on complex, long-running tasks and agent workflows” is also not a shocker as “agent” and “agentic” are The hot new AI buzzwords for 2025, and again Claude was an early entrant in this area with its Computer Use released in October of last year.
I’ve been using Claude 4 since it hit the Play Store and desktop app, so for a week now. All of my interactions with it have been good. I only used the word “solid” in the title, because I’m used to Claude being consistently good, and often better than good.
Here are some of the things I’ve tried it out with and been impressed with:
Summarizing a lengthy Cambridge journal article on climate change and cybersecurity, and how they are now linked
Doing a very good job at taking a list of specific Sysmon events (a type of security event that has sometimes been said to be a way to “always find a thing”, as in a bad thing, and how to find those events, with proper search syntax, in Splunk
A stellar job adding a section to my post at Cyber Sherpas on the Cyber Kill Chain.
And best of all producing a script to automate putting together a cyber threat alerts brief for me.
Claude has also added Voice Mode sometime over the last week or so. I’ve only tried this out on a few occasions so far, and based on those, I’d say it needs a little polish. It feels slow to launch than ChatGPT advanced voice mode. It also looks like it is not yet continuous, as ChatGPT’s is. It prompts to “tap here to send”. I expect this will be getting some updates soon.