@tjrobertson52: GPT 5.5 is the real deal. But should you switch from Claude for work? Here's what actually matters 👇 #ChatGPT #AI #Claude #SmallBusiness
Use both if possible. I use them as a checks and balances approach.
2026-04-24 23:49:17
24
paniardo :
Deepseek dropped theirs and so far benchmark either matches or slightly exceeds gpt 5.5 at 1/20th cost. For writing task, I want to test on deepseek and finetune them on Claude. It's better to test for our use case instead of relying on benchmarks alone.
2026-04-25 00:34:37
1
ecocawphezj :
How is ChatGPT 5.5 at writing? I canceled my account after they stopped letting us use 4.2, because I found that the early versions of five couldn't produce very good writing, and switched to Claude which is super expensive. maybe it's time to switch back?
2026-04-25 06:44:49
4
SalmonToro94 :
Use model selection policies. You’ll thank me later. Codex does it automatically if you set up the policies. Claude code will prompt you to change models based on this model policy selection.
2026-04-27 02:31:56
2
Wanky Doodleㅤㅤㅤㅤㅤㅤㅤㅤㅤㅤㅤㅤㅤㅤㅤㅤㅤ༽ :
It's Claude and Gemini. ChatGPT is so far behind, it's not even part of the race anymore.
2026-04-25 05:27:55
3
John McLaughlin :
Anthropic built the infrastructure around the harness and the LLM where you have to use their ecosystem to build what you want to build. The infrastructure and UX make it much easier to use and build with than Codex. As other models improve, their infrastructure is becoming their moat. Though, they’ve noticeably been pinching usage limits and it’s getting quite expensive to build with. Unlike OpenAI where you get wayyy more use out of their models than Opus.
2026-04-25 17:48:47
2
Brian 🇺🇸 :
I use different models in OpenCode for different tasks. Honestly GPT-5.4 is better than Opus 4.7 for planning code architecture.
2026-04-25 01:31:48
5
Eli Peters :
claude code for code... chatgpt to test claude... cause claude can hallucinate. you have to dev to get the best of claude
2026-04-25 07:43:11
1
Gengxi.AI :
So?
2026-04-28 01:45:48
4
Franck :
how could I get more information on the prop pose and use of the documentation you've mentioned? in pretty much clueless so far
2026-04-25 15:33:23
0
Nicola Torcutti :
Are you considering for your clients use the wiki knowlage base with obsidian by karpathy?
2026-04-25 07:19:44
1
༺😈♛ Ðв ♛😈༻ :
I am new to this and doing exactly this. I work on all Ai using same project folder getting better results and keeping the track of progress by package folder. Every tweak get a new package-Ai Name-ver# and once I approve the change it update the MD file.
2026-04-25 23:06:46
3
okkun オックン :
does chatgpt 5.5 have skills like Claude skills
2026-04-25 07:52:22
0
blackhatseo :
5.5 still sucks in real world use vs opus 4.7
2026-05-06 11:58:07
1
Gether.To :
Very interesting point about documentation.
I would add an MCP server over your proprietary data and you are pretty much agnostic to the agent you use.
2026-04-25 10:15:47
1
Snowcrash :
Gotta have both. It’s the only way.
2026-04-27 01:27:31
1
John Helldiver :
Sop?
2026-04-25 13:55:25
0
Dr Tony | Author & Mind Coach :
I added a code bug to both Chat and Claude and showed Chat the patch Claude made, its reply highlighted all the issues it missed. I’m sticking to Claude for now!
2026-04-25 05:34:19
3
Fero_Za :
If you're only using one or the other, you are way behind. Cancel your netflix and get both
2026-04-25 06:40:07
2
Pmakk :
The images are freakin amaaaaazing!!!
2026-04-25 04:00:09
2
N0_Vaseline :
Switch from GPT to Claude MAX + Obsidian + Github systems and processes are smooth and powerful. Don’t see how GPT can match
2026-04-25 01:23:40
1
maris_nekretnine_pula :
Nice can you help me to preparong documentstuon? Or can you send me a video? Thx 👏
2026-04-25 05:15:38
0
Ed Enjury :
it's like an Apple or Android question for phones in my opinion. Both do the same thing ... although these AI models are a similiar price, unlike the phone hardware
2026-04-25 06:01:07
1
liked your comment :
Claude in excel and Claude in PowerPoint still wins, no question.
2026-04-25 13:41:49
1
Robert Grzesik :
5.5 is fantastic
2026-04-25 15:33:19
1
To see more videos from user @tjrobertson52, please go to the Tikwm
homepage.