Claude Code and Claude 3.7 Sonnet - Our most intelligent model to date
by•
It's a hybrid reasoning model, producing near-instant responses or extended, step-by-step thinking. One model, two ways to think.
We’re also releasing an agentic coding tool: Claude Code.
Replies
Dune
Been watching anthropic improve claude for a while but 3.7 sonnet feels like a real leap. the reasoning on complex multi-file tasks is noticeably sharper and claude code makes it so easy to point it at your actual codebase rather then pasting snippets into a chat window. the combo of a smarter model with a proper agentic coding tool is exactly what i've been waiting for, congrats to the team 🔥
Been using Claude 3.7 Sonnet heavily for the past few months
while building my first SaaS solo — the extended thinking mode
is genuinely different from anything else I've tried for working
through complex architecture decisions.
The jump from "here's code that works" to "here's why this
approach is better than the three alternatives" changed how I
use it day to day.
Question for the team: is Claude Code designed more for
greenfield projects or does it handle large existing codebases
well? Curious how it manages context across 50+ files.
It actually reads your codebase and understands your patterns instead of just autocompleting. Described a feature in plain english yesterday and it wired it into my existing architecture perfectly. The 200k context window is a game-changer for larger codebases
Honestly didn't expect to get this hooked. After weaving Claude into my daily workflow, going back feels genuinely painful.
I'm not a developer — so for me, understanding the problem matters more than just spitting out code. That's where Opus really stands out. Compared to ChatGPT o3, the way it grasps context and intent is on another level. It doesn't just answer what you asked — it gets what you meant. Btw, hallucination is still the elephant in the room. lol
it is great for my daily work, but it costs more
the cache TTL change that just surfaced on HN is worth watching in context of this launch. when Anthropic shortened cache from 1h to 5m, it directly impacted long Claude Code sessions — the system prompt gets re-processed more often, costing more tokens and adding latency. for teams running agents in CI or as daemons, this is a real operational cost. the 3.7 model improvements are great, but the infrastructure layer (caching, billing transparency, permission auditing) is where the next round of trust gets built.
Claude Code provides access to the social elevator. Is changing the world as we know it and the people out of IT dont know it. Love this AI which help you create each line of code you are not able to do it. To be honest I dont see much difference with 3.6 but I can say that this ai is almost perfect to code. Truly recommended.