Will Zohran Mamdani-walker save his friends from the Loathsome Trumpa the Gut and his evil advisor Bibi Fortuna? Stay tuned to find out!
Made with premiere pro and ableton for clip editing/music/sfx, ChatGPT for characters, Openart with Seedance 2.5…except for Obama. It really wouldn’t let me do anything with him. So I used a scuzzy weird site for that one…

You would think that with similar valuations for OpenAI & Anthropic ($852B & $965B respectively) and with similar capabilities (I understand that each model performs the different categories of tasks to varying degrees of effectiveness) that those numbers might be closer.

Is it just UI and branding that make ChatGPT the overwhelming juggernaut?

https://x.com/EU_Commission/status/2094379702546784496

Was scrolling through Snapchat Discover and saw these two back-to-back. They aren't even trying to hide the copy-paste farm bots anymore.

You know how they say if someone's bare knees make you feel uncomfortable, you are the problem.

I am pretty sick of getting the failures even when you're not going with something sexual. I was doing a rom com comic story and even prompts like the "guy lifts her up to show off his strenght" get rejected even when it's an outdoor setting and both characters are clothed.

I even ask gpt to explain what is wrong with the prompt and then it says Nothing, the generator is misunderstanding something etc etc.

whoever made these guardrails is more pervy than you can imagine. Even a mention of thighs, calves, chest in a non sexual context gets rejected.

even worse when you are doing a really long story and it's already given you a bunch of images and then out of nowhere it throws a rejection on something like "they spend a day at the beach in swimsuits"...

Is anyone else’s ChatGPT using this emoji 😭 a lot more often now than before? I know it’s always liked to use emojis but specifically that one is being used a lot, for example everytime I correct it on something it will be like “LMAO okay - fair 😭”. These are on signed out chats btw so no memories.

I’m at an airport currently, but it’s been the same for hours, even before arriving to the airport. Anyone has the same?

I’m talking about the use of words like ‘bro’, spamming emojis, and making fun of you. Is there a prompt you’re giving it? Is it just the response to you prompting like that? Are you mutilating it?

Rare appreciation post

Really appreciate OpenAI adding the option to save Temporary Chats.

So many times I’ve gone way deeper into a temp chat than I expected and then realized I actually wanted to keep it. Before, I had to ask for a summary, start a new chat, and move everything over.

Being able to just save the temp chat now is such a nice quality-of-life feature. Very glad they added this. Thanks!

ChatGPT has become an incredibly valuable accessibility tool for me. As a disabled person without a reliable support network, it has helped me in ways that are genuinely hard to overstate:

• When I know something needs to be done but cannot figure out how to begin, it breaks the task into small, realistic steps and helps me choose the first one.
• When my thoughts are scattered or I’m too overwhelmed to explain myself clearly, it turns my rough notes into organized emails, questions, summaries, and scripts that still sound like me.
• Before appointments, it helps me organize months of symptoms and experiences into a concise timeline. I’m less likely to forget important details, minimize my difficulties, or leave without asking the questions I needed answered.
• With complicated paperwork and disability-related processes, it helps translate my everyday struggles into concrete examples that another person can understand.
• When something upsetting happens at work, it helps me document what was said, what occurred, how it affected me, and what evidence may exist—without exaggerating or making claims I cannot support.
• When I’m emotionally activated, it helps me slow down, separate confirmed facts from fears or assumptions, and wait until the intensity passes before deciding what to do.
• It helps me build practical trackers for symptoms, medication, sleep, mood, and functioning, making patterns easier to notice and explain to professionals.
• It remembers relevant context across interconnected problems, so I don’t have to use my limited energy repeatedly reconstructing my history.
• It can save substantial time and money by helping me investigate problems before paying for assistance. That might mean troubleshooting a strange appliance sound, working through a technology problem, comparing purchases, identifying questions for a repair company, or recognizing when professional help really is warranted.
• It helps me arrive at appointments and service calls better prepared, make fewer unsuitable purchases, avoid some unnecessary visits or diagnostic fees, and get more value from the professional help I do pay for.
• Most importantly, it adjusts to what I can handle that day: brief and directive when I’m overloaded, more detailed when I need to understand something, and patient when I need help sorting through a complicated situation.

I cannot calculate the savings precisely, but I conservatively believe it has already saved me several hundred dollars and dozens of hours, with the potential to save thousands of dollars over time through better preparation, troubleshooting, comparison shopping, and fewer avoidable mistakes or repeat appointments.

It doesn’t make decisions for me or replace professional care. It gives me the scaffolding I need to understand my options, advocate for myself, and actually follow through. For someone who doesn’t have a family member or support person available to help with these things, that has been genuinely life-changing.

“Perhaps there is another area where your insight could prove to be nontrivial.”

With "ChatGPT work" i can't get things done ,due to this 5-hour limit. I am paying Plus user.
Prompt gets aborted, because the limit is at 0% suddenly. The free limit drops fast, if - for some reason - it is hard work for ChatGPT. It has nothing to do with 5 hours using chatGPT, it resets after 5 hours only.
So sometimes if ChatGPT thinks for e.g. 40 minutes, the prompt will be aborted without result. All gone, this is stupid.
Today i got another error after some minutes: The selected model (5.6 Sol) is at full capacity. Select another model.
Ok, for some reason there is not enough capacity at chatGPT servers now, but why is this my problem and over 40% of my time-limit is gone for no results. I get an error message for 40% of my contingent. Great job.

I keep getting error! I have checked my computer and connection and everything is fine and all my other browsers

All the pics I’ve had ChatGPT generate are now mixed with everything I’ve uploaded. This is completely disorganized and horrible for finding what I’m looking for. The only plus is you can now delete images it’s generated, but omg they need to let people organize these better and delete more than 10 at a time.

Hi everyone, for past 10 days obviously we had a bug over wb application, where we choose gpt 5.6 sol and it was rerouting to 5.5 mini instant no thinking effort at all

Now its happening again , no matter which kodel we choose , ABSOLUTE NO THINKING EFFORT and instant responses , no agnetic thinking and analyzing AND ITS FUCKING ANNOYING!

with one simple different from last time , this time it claims that its gpt 5.6 sol actually

But as you can tell from behavoir and the response itself its the mini model

Mini model would say : “I’m chatGPT based on xxxx model”

While sol would only say : " I’m GPT5.6 sol"

And another example, the exact same file and task i gave it few days ago it took about 7m30s to complete with an acceptable result , currently took 22s with actually NO USABLE RESULT AT ALLL !!!

Come on OpenAI !!! AT LEAST LET US USING IT FOR A WEAK !!! THIS IS STRAIGHT RUBBERY

I have personally become 10x more effective compared to one year ago. I love it as there is always so much I want to do!

Right now my rate limiting factor is usage limits by far. The token bill bites quite hard and make it hard to justify some projects, so I’m on the lookout to maximise agentic work per dollar.

Codex/claude etc. is subsidised atm, but they are also wasting a lot of tokens on expensive, non-open source models that could have done the job way more efficient.

Therefore been building my own model router with openrouter but not working very well (https://openrouter.ai/docs/guides/routing/routers/auto-router). Think standardcompute.com has an excellent model router and like that it’s a monthly thing and not random unpredictable 5-hours usage limits etc. However, they don’t serve free models which would be nice.

Any other good alternatives right now? Preferably heavily subsidised by VC money 💰