Anthropic Claude 3.5 and 3.7 have been the leading models for coding. They were being threatened by Google Gemini 2.5 but now Claude 4 Sonnet and Opus are out. Cloude 4 Sonnet and Opus are next level for coding.
Claude 4 will ask more questions to be certain you and it know what you want to code and what should it code.
It will then code and create answers with better analysis.
Agentic coding and Agentic terminal coding are substantially better.
Agentic coding is at 80% versus 64-69% for competitors.
Agentic terminal coding is at 50% instead of 30% for competitors.
A second year AI university student did testing just now and Claude 4 Sonnet solved problems that 3.7 and Gemini 2.5 Pro could not. It then refined that solution based upon descriptions of the desired improvements.
Early testers said it could code autonomously for up to seven hours.
Clause 4 refactored an entire codebase (50k lines) from Vite/React to Turbopack/Next.js with barely any oversight, in less than 60 minutes.
Introducing the next generation: Claude Opus 4 and Claude Sonnet 4.
Claude Opus 4 is our most powerful model yet, and the world’s best coding model.
Claude Sonnet 4 is a significant upgrade from its predecessor, delivering superior coding and reasoning. pic.twitter.com/MJtczIvGE9
— Anthropic (@AnthropicAI) May 22, 2025
Claude 4 just launched from @AnthropicAI
I was lucky to get early access.
tl:dr from my experience:
1. It's still best in class at writing and editing
2. It's just as good at coding as Gemini 2.5It built this full working version of Tetris in one shot – link to play below: pic.twitter.com/LXTSUcfoWv
— Peter Yang (@petergyang) May 22, 2025
sonnet 4 is also a work of art, especially when used in an agentic harness like claude code
i asked it: "can you create a manim video that demonstrates how black holes work? make use of 3D animations"
this was the one-shot result pic.twitter.com/IrjBSn4XnC
— zack (@wenquai) May 22, 2025
Claude 4 Opus coding non-stop for 7 hours?!
With any amount of reliability and throughput, this is a step change in capability. 🤯 pic.twitter.com/CYCcSes1x0
— Justin Halford (@Justin_Halford_) May 22, 2025
Brian Wang is a Futurist Thought Leader and a popular Science blogger with 1 million readers per month. His blog Nextbigfuture.com is ranked #1 Science News Blog. It covers many disruptive technology and trends including Space, Robotics, Artificial Intelligence, Medicine, Anti-aging Biotechnology, and Nanotechnology.
Known for identifying cutting edge technologies, he is currently a Co-Founder of a startup and fundraiser for high potential early-stage companies. He is the Head of Research for Allocations for deep technology investments and an Angel Investor at Space Angels.
A frequent speaker at corporations, he has been a TEDx speaker, a Singularity University speaker and guest at numerous interviews for radio and podcasts. He is open to public speaking and advising engagements.
I just tried to evaluate Claude 4 by subjecting it to a coding problem I’m working with together with GROK 3 (which seems to suck a parsing, shell scripting and Python refactoring).
I submitted a description of the problem I use to transfer the knowledge between sessions (that crash now and then) plus two small source code files (~35KB worth).
I didn’t even get to the data input files or examples of output needed for debugging.
This was enough to make Claude 4 choke and throw an error stating too much input for the session. This means I can’t even evaluate the model for one iteration. I can’t even present the scope of the problem 🙁
It will not matter if I get 5 times more with the pro subscription. It will still be useless.
With GROK, I can keep iterating and sending stuff all day long for 2 – 5 days before it chokes and I have to start fresh.
Anyone had better luck?
Anthropic’s Claude is great, however when I went to pay to use the service, I was ripped off — even though I show as having paid, I’m also still only showing up as a free user and the API is hard limited for me.
Worse still their support has ignored my requests for help and won’t give a refund (a lot of money for me too) … so my two bits, avoid anthropic’s paid service like the plague.