Post

@CodeWithAmann I see people choosing a third option
I'd take:
- ollama pro for 20$ for insane oss models limits
- ChatGPT for imagen and codex, 20$ goes a long way
If for some reason ollama pro isn't enough, I'd drop Codex and go ollama pro for 100$
I just like the versatility
English

@CodeWithAmann Hey @grok count who has the most recommendations this thread Codex, Claude, or third option?
@CodeWithAmann you thank me later
English

@CodeWithAmann If I had $100 I’d go with Claude Code right now because it has better long coding sessions, cleaner outputs & feels more like a senior dev beside you.But Codex is catching up fast with automation + desktop workflows.
English

@CodeWithAmann Limits are the only problem with Claude rn, so Codex.
If I just want Claude at any cost, I'll get Claude Code and put 20$ into DeepSeek credits.
Deepseek v4 Pro as my workhorse and Claude Opus 4.7 for the most complex tasks.
English

@CodeWithAmann codex pro. opus 4.7 is incredible but anthropic nerfed the limits so hard you end up stuck on gemini 3.1 pro half the day. openai's limits are just the sweet spot for actual vibecoding.
English

@CodeWithAmann Claude. I definitely had the suspicion that their explosive growth and the limits they placed was due to a lack of compute. I trust the org to do what they need to do. Although in my case I have both and paying for the 20x on each
English

@CodeWithAmann Go for Claude. GPT is getting worse day by day
English

@CodeWithAmann I’m vibing on Antigravity, Cursor, Codex, and Claude — sometimes I’m wondering when I’d be able to choose the one that suits me the best.
I decided to use all to see which one performs better, now each is used for particular task or project.
Most used to the credit limits
English

@CodeWithAmann Codex this month, probably claude next month. /rinse and repeat
English

@CodeWithAmann Claude Code for sure. It understands your entire codebase, not just a single file. For real projects that’s a game changer. Codex is good but Claude Code just thinks deeper 🔥
English

Tough one man. 😅
I’m team Codex for now.
That GPT-5.3-Codex-Spark sounds perfect for the daily grind + the double usage promo till end of May is actually useful instead of marketing fluff.
Claude Max looks tempting with the 20x limits and early features but I’ve gotten wrecked by Anthropic’s rate limits before when I’m in flow state.
Anyone actually using both heavy?
Curious what’s winning for real coding hours.
English

@CodeWithAmann If you are working on gui and won't be running a few sessions in parallel then Claude without any exception.
Claude's usage limit is the only thing bad about it when it comes to coding. Go with codex if usage matters for
you.
English

@CodeWithAmann If you are a coder and dont use Claude and instead use GPT, then I feel sorry for you lol
English

@CodeWithAmann Claude Code for development, and Codex for design!
English

@CodeWithAmann Claude Code 100% it builds better but it has lots of bugs sometimes but it’s nice
English

@CodeWithAmann Codex ...opus 4.7 did not help me debug..codex with 5.5 did it in one go
English

@CodeWithAmann depends
want speed = codex
want therapy + occasional code = claude
English

@CodeWithAmann Codex, without a doubt. So much more for your buck.
English














