Renzo Canepa

@signulll This is really awesome. Do you know when they are releasing it to all customers? I can't see OpenAI models on BedRock at the moment.

ppl underestimate how big of a deal it is that openai models are now on bedrock. i’ve met so many co’s that defaulted to anthropic & claude because they were already on bedrock, & for a long time that was basically the path of least resistance. this is huge for openai model accessibility.

GPT Image 2 is so good! It can create amazing infographics with unbelievably accuracy with a very simple prompt.

@seraleev Puerto Varas!

I love this little town on a lake with a volcano view. Ran the 21k here a few times. You run, you enjoy the scenery. Highly recommend if you ever come to Chile.

@rezoundous The decrease in token usage should more than compensate for the higher price, I think. That said, I see my weekly limits decreasing faster than with GPT-5.4.

GPT-5.5 isn't mind blowing or anything, but I feel it's a good step up from the 5.4. But I definitely feel it shouldn't be twice as expensive as 5.4.

@enjojoyy Good decision. GPT-5.4 and GPT-5.5 are awesome models. I still pair them with Opus 4.7; but GPT models have the last say always (on coding tasks).

I just cancelled my Claude Code subscription to switch to Codex

@vikrambuilds Codex for iterative coding, complex tasks, existing codebases. Claude to one-shot UI or apps from scratch. If I had to pick only one, it would be Codex.

Claude or codex?

@RoundtableSpace GPT-5.5 feels slightly superior to 5.4 and much faster. However, Opus 4.7 feels like the same thing or even worse at times.

Which one are you more impressed with, GPT-5.5 or Opus 4.7?

@dramaricic I guess it comes down to personal taste. I prefer GPT-5.4 over Opus 4.7 every day. I use CC models for niche tasks and reviews (GPT is still the main reviewer). So, for me, GPT-5.5 is definitely superior to Opus 4.7.

Anyone tried GPT-5.5 yet? Is it comparable with Opus 4.7?

@om_patel5 This looks awesome! How did you create the video?

THE BEST REDDIT MARKETING TOOL FOR THE FASTEST DISTRIBUTION ON YOUR STARTUP JUST LAUNCHED reddit is the fastest way to grow a startup and get paying customers right now every other platform is packed with creators trying to SELL you something reddit is packed with BUYERS publicly asking for a product to solve their exact problem the top post/comment in almost every subreddit is "does anyone know a tool that does X" in different niches that's a customer literally typing out the problem YOU SOLVE and waiting for someone to answer so I built a tool that does the whole thing for you: > scans every subreddit your buyers hang out in and finds people with real buying intent > writes content that matches the tone of each subreddit so mods don't ban you > ranks your posts on google, chatgpt, perplexity, and claude so one post drives traffic for months (GSEO + SEO) > gives you keywords scored on conversion potential, not just search volume > FULL MCP support on every AI client (claude/claude code, cursor, vscode, etc) went from 0 to 10k MRR just by showing up in the right reddit threads at the right time + making reddit content that ranks on every keyword no other platform gets you paying customers this fast watch me use it in under 2 minutes:

@aniketapanjwani Do you explicitly ask GPT to generate designs for coding tasks, or how does GPT-Image-2 makes the coding experience with regard to UI/UX design?

Until yday there were only two things I found CC better at than Codex - UI/UX design - writing Since ChatGPT Images 2.0, from my early testing, Codex now beats out CC for design - even with CC's design update. I've also noticed - similarly to other observers - recent very stupid behavior by Opus 4.7 for writing. I'd prefer to use one tool for everything, so I'm going to be pushing at Codex for writing the next few days to see if I can match CC quality

Massive update to hummingbytes.com 🚀 Seedance 2.0 and GPT-Image-2 are officially live—bringing the leading models in video and image generation to your toolkit. Want to try them out? Hit me up for a discount code and start building your next project today. 💎

Google is officially playing catch-up. 📉 Veo 3.1 is no longer the video king (Seedance 2.0 has the lead). Gemini 3.1 Pro is consistently outcoded by Anthropic and OpenAI models. And after a very long period at the top, Nano Banana Pro has lost the image throne to GPT-Image-2.

@mark_k Yes; I went back to using the API :)

@rcanepag It's totally broken

I wanted to access GPT-Image-2 from the OpenAI API. It told me I need to "Verify" my account first with government ID and camera. So I went through all that bullshit (why even?), only to be greeted with this: "No eligible models available". Do you hate your customers, OpenAI 🤦‍♂️

@Jashanx_gill I would say GPT-5.4 for coding :)

Perplexity - for research ChatGPT - for images Claude - for coding What else to add?

@mark_k Oh, I just realized I can't select the model anymore. I guess it is a bug.

@mark_k Is this due to your plan?

@Azure Every generation request takes so long that the service is unusable. Even 1024x1024 images can take more than 5 minutes to complete.

OpenAI's GPT-image-2 is coming to Microsoft Foundry with notable updates for production image workflows: improved editing performance, enhancements to image quality, new multilingual and thinking capabilities, and broader support for resolutions and aspect ratios—including 4K. Together, these updates give teams more flexibility when generating and refining images at scale. Learn more: msft.it/6016v82XA

@rickdev_ai @pigeon__s Where did you find this table?

@pigeon__s It’s not that cheaper

GPT-IMAGE-2 IS ALSO WAY CHEAPER THAN EVEN GEMINI ***FLASH*** IMAGE 😭 THIS IS SIMPLY THE MOST PATHETIC MOGGING IVE EVER WITNESSED

@bridgemindai GPT-5.4 is already beating Opus 4.7, isn't it? Since the jump from Opus 4.6 to 4.7 was so mediocre, GPT-5.5 is likely going to blow it out of the water.

GPT 5.5 is going to beat Claude Opus 4.7. I'm calling it now. Anthropic did not set the bar very high. The improvement from Opus 4.6 to 4.7 is real but incremental. Benchmaxed benchmarks. 35% more tokens. Reduced security. Sam Altman said GPT 5.5 Spud is going to be a leap. Two years of research. Not an incremental update. He better be telling the truth. I used to love OpenAI. Then Claude Code happened and I switched. Now Anthropic is rate limiting me on two $200/month subscriptions. Testing GPT 5.5 the moment it drops tomorrow. BridgeBench results same day.

@burkov In my experience, using an iterative approach can work pretty well. However, if you try to one-shot a mildly complex UI it will probably fail in a really bad way.

Codex is absolutely hopeless with UIs. I haven't seen such dumbness since o1.