Chat GPT & Generative AI tools
-
Big Bad Barney
- Posts: 397
- Joined: Mon Feb 04, 2019 6:00 am
Hi Claude user, Codex/ChatGPT doesn't have those commands anymore....at least not in their app, perhaps maybe on the CLI....it's annoying
uhhmmm I did check if those worked in codex and google ai says:
If you are using OpenAI's Codex CLI, OpenAI's desktop terminal integrations, or other developer-focused AI toolkits, /compact and /clear are fully supported commands, but yes via web : ChatGPT does not have built-in native slash commands equivalent to /compact or /clear (typo of /clear) found in terminal-based coding assistants like Claude Code.
Maybe this helps ?
Equivalent Actions in ChatGPTEquivalent to /compact (Summarizing context): ChatGPT has no automatic conversation compression command. You must manually prompt the model by typing "Summarize our conversation so far and focus only on..." or start a fresh thread.
Equivalent to /clear (Wiping the screen/history): There is no /clear command in the web chat interface. To start completely fresh, you must click New Chat in the upper-left corner or bulk-delete your history via Settings > Data Controls > Delete all chats.
If you are using OpenAI's Codex CLI, OpenAI's desktop terminal integrations, or other developer-focused AI toolkits, /compact and /clear are fully supported commands, but yes via web : ChatGPT does not have built-in native slash commands equivalent to /compact or /clear (typo of /clear) found in terminal-based coding assistants like Claude Code.
Maybe this helps ?
Equivalent Actions in ChatGPTEquivalent to /compact (Summarizing context): ChatGPT has no automatic conversation compression command. You must manually prompt the model by typing "Summarize our conversation so far and focus only on..." or start a fresh thread.
Equivalent to /clear (Wiping the screen/history): There is no /clear command in the web chat interface. To start completely fresh, you must click New Chat in the upper-left corner or bulk-delete your history via Settings > Data Controls > Delete all chats.
I think once you install Codex (or Claude Code) and connect your ChatGPT account, you'll never want to work over the web again. With Codex, you can work directly on existing folders, documents, databases, etc. — it's a hundred times more productive. As mentioned before, make sure to spend time following best practices to safeguard against errors the AI tool might make, etc."
-
Big Bad Barney
- Posts: 397
- Joined: Mon Feb 04, 2019 6:00 am
yeah agree, desktop is where it's at.....I been using the Codex App on windows the past few months. It doesn't have the /clear or /compact. Before that I was using CLI as I was mixing Claude and Codex stuff, but my memory is shit so I can't remember ..... I kind of like Opencode as an app...But just easier/better to stick to one company. For me that's been OpenAI lately... Anthropic lost me around Opus 4.7 and GPT5.5 was pwning it and I haven't gone back to try 4.8 or 5 or anything....
6 or 12 months back Anthropic had a promo to get people to use their new web interface...I figured, oh yeah, web sucks but...I'll give it a crack for free tokens....hammered it for 2 days and got banned....lol...
..... I kind of like Opencode as an app...But just easier/better to stick to one company. For me that's been OpenAI lately... Anthropic lost me around Opus 4.7 and GPT5.5 was pwning it and I haven't gone back to try 4.8 or 5 or anything....6 or 12 months back Anthropic had a promo to get people to use their new web interface...I figured, oh yeah, web sucks but...I'll give it a crack for free tokens....hammered it for 2 days and got banned....lol...
-
firlandsfarm
- Posts: 3688
- Joined: Sat May 03, 2014 8:20 am
Tokens. I know what they are for Claude ... they are the most frustrating monitoring of use of AI ever invented (well on Claude they are). First time I came across them was when Claude was in the middle of preparing a long document for me and suddenly stopped saying it had run out of tokens and that because of that I couldn't ask it to summarise the chat to kick-start a follow-on chat! 
So I started a new chat discussing what the hell tokens were and we agreed it would put in my profile that in future it should advise me when I get to 75% used as a warning and 90% used as a flag to get a summary and move on. Great I thought. But the first time it should have been tested it didn't warn me. I reminded it of the agreement and it basically said it no can do! I copied and pasted the conversation around the 'agreement' when it said I was lucky to get a connection that could! WTF! I asked it to explain and was told different connections can preform different internal tasks! Well that was enough for me, I switched back to GPT. But my reason for explaining that is I have had longer chats on GPT than Claude and never had a Token issue. Indeed, I've only just come across Tokens on GPT. What are they fo?
So I started a new chat discussing what the hell tokens were and we agreed it would put in my profile that in future it should advise me when I get to 75% used as a warning and 90% used as a flag to get a summary and move on. Great I thought. But the first time it should have been tested it didn't warn me. I reminded it of the agreement and it basically said it no can do! I copied and pasted the conversation around the 'agreement' when it said I was lucky to get a connection that could! WTF! I asked it to explain and was told different connections can preform different internal tasks! Well that was enough for me, I switched back to GPT. But my reason for explaining that is I have had longer chats on GPT than Claude and never had a Token issue. Indeed, I've only just come across Tokens on GPT. What are they fo?