惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

Martin Fowler
Martin Fowler
T
The Blog of Author Tim Ferriss
J
Java Code Geeks
M
MIT News - Artificial intelligence
F
Fortinet All Blogs
Cyber Security Advisories - MS-ISAC
Cyber Security Advisories - MS-ISAC
B
Blog
Microsoft Azure Blog
Microsoft Azure Blog
I
InfoQ
Microsoft Security Blog
Microsoft Security Blog
N
Netflix TechBlog - Medium
G
Google Developers Blog
L
LangChain Blog
腾讯CDC
大猫的无限游戏
大猫的无限游戏
U
Unit 42
Google DeepMind News
Google DeepMind News
人人都是产品经理
人人都是产品经理
罗磊的独立博客
钛媒体:引领未来商业与生活新知
钛媒体:引领未来商业与生活新知
小众软件
小众软件
The GitHub Blog
The GitHub Blog
博客园_首页
GbyAI
GbyAI

DEV Community

Authentication Security Deep Dive: From Brute Force to Salted Hashing (With Java Examples) Why AI Systems Don’t Fail — They Drift Spilling beans for how i learn for exam😁"Reinforcement Learning Cheat Sheet" I Replaced Chrome with Safari for AI Browser Automation. Here's What Broke (and What Finally Worked) How Python Borrows Other People's Work The $40 Architecture: Processing 1 Billion API Requests with 99.99% Uptime Vibe Coding: A Workflow Guide (From Zero to SaaS) Most webhook security guides protect the wrong side. The scary part is delivery. Headless CMS for TanStack Start: Build a Blog with Cosmic EU Age Verification App "Hacked in 2 Minutes" — What Actually Happened Comfy Cloud’s delete function does not actually remove files Running AI Models on GPU Cloud Servers: A Beginner Guide Event-driven media intelligence with AWS Step Functions and Bedrock I scored 500 AI prompts across 8 quality dimensions — here's what broke How to Call Google Gemini API from Next.js (Free Tier, No Backend Needed) The Portal Protocol: Reclaiming Human Connection in the Age of AI How to Fix Your Team's Scattered Knowledge Problem With a Self-Hosted Forum Intro to tc Cloud Functors: A Graph-First Mental Model for the Modern Cloud Designing Multi-Tenant Backends With Both Ownership and Team Access I Built a Neumorphic CSS Library with 77+ Components — Here's What I Learned PostgreSQL Performance Optimization: Why Connection Pooling Is Critical at Scale Cómo construí un SaaS multi-rubro para gestionar expensas en Argentina con FastAPI + Vue 3 🚀 I Built an Ethical Hacking Scanner Tool – Open Source Project I Replaced /usage and /context in Claude Code With a Single Statusline A Pythonic Way to Handle Emails (IMAP/SMTP) with Auto-Discovery and AI-Ready Design I Collected 8.9 Million Polymarket Price Points — Here's What I Found About How Markets Really Move EcoTrack AI — Carbon Footprint Tracker & Dashboard Everyone's Using AI. No One Agrees How. 5 self-hosted ebook managers worth trying in 2026 Building Your First AI Agent with LangChain: From Chatbot to Autonomous Assistant
I Was About to Cancel Claude. Now Gemini Is Rate-Limiting...
MediBlackSand · 2026-06-25 · via DEV Community

The Almost-Cancellation

About six months ago I was one renewal away from cancelling Claude. Not because it got worse, because I just wasn't opening it. Rate limits I kept hitting on the rare day I did, and a Gemini 3 Pro that had just landed and felt better at almost everything I cared about: Python, JavaScript for the artistic coding side projects (a whole separate post, another day), general "what can this thing actually do" exploring.

Gemini 3 Pro earned the roughly 30 AUD a month on its own. Add the Drive storage I was already half-paying for elsewhere and the image and video generation that was genuinely solid, not a tacked-on afterthought, and the upgrade looked obvious. I downgraded Claude in my head before I'd even opened the app to do it.

Then Something Shifted

Sometime this year that stopped being true. Not Gemini getting worse, Claude getting sharply better: coding accuracy, front-end output, MCP, agent behavior, all of it improving in a way I noticed week to week instead of release to release.

I started running the same prompts through both, mostly out of curiosity at first. Then out of habit, because I kept trusting Claude's answer more often than not. Same task, same wording, and Claude's version was the one I'd actually ship.

The Downgrade

So I went the other way on Gemini. Dropped down to Google AI Plus, the cheap tier, around 15 AUD a month. I wasn't using it daily anymore, and I didn't want to keep paying full price for a tool that mostly sat there.

These days Gemini is what I open for googling instead of the traditional "googling", or for the smaller stuff I don't want eating into my Claude usage. Nothing heavy. Nothing I'd call real work.

And even at that, low-stakes level, I'm getting a lot more "I'm having a hard time fulfilling your request, can I help you with something else instead?" on requests that aren't unusual at all. Just now, checking back in to write this, I've lost access to the better model completely and landed in a flat rate limit. Not slower. Just off.

What This Actually Is

I'm not saying Gemini is bad. Six months ago I'd have written the exact opposite post about Claude, and meant it just as much. What I'm noticing is smaller and more annoying than "which one wins": 15 AUD a month bought me a tool I can poke at, not one I can actually lean on. That's not really a capability problem. That's the cheap tier telling you, politely, that you're not the priority.

Closing Thought

I didn't downgrade because Gemini got worse. I downgraded because I wasn't using it enough to justify the higher plan. But the cheap tier doesn't meet you at "occasional and light," it meets you at "barely worth keeping." Pay less and you don't just get less model, you get less patience for using it at all.