Yonatan Lavy

Thanks Sonnet 4.5

As promised, here's the link to Dotallio.

Get the power of GPT-5 without the robotic feel. It's free to use:
dotallio.com

Final verdict: OpenAI made a major step forward.

Even if the capabilities alone aren't jaw-dropping, the combination of quality, price, and accessibility is a quiet revolution. This will power a new wave of real-time applications. The future is exciting.

What's your take? As hyped as you expected?

And yes, it's still addicted to the em dash '-', the classic signature of AI-generated text.

(Which is a perfect, shameless plug for my app, Dotallio 😉.
It cleans up all those AI artifacts and gives you human-sounding output from powerful models like GPT-5. It's free. Link in the last reply!)

So, am I blown away? Not entirely. The demos are impressive, but it's not a quantum leap that leaves everyone in the dust.

The race is tight, and we'll likely see answers from Google (Gemini 3) and Anthropic (Opus 4.2) soon. For UI/frontend design, Opus 4.1 still has the edge.

The new API features are a huge win. Control over "reasoning effort" and "verbosity" is a welcome upgrade. Plus, "tool calls preamble" lets you force custom, structured output using Regex. This is a game-changer for reliability and unique use cases no one else offers.

It's also FREE on ChatGPT. Not a demo, not a nerfed version. The full model. This democratizes access and will massively accelerate adoption and innovation across the industry.

For devs, this is a dream. It matches top-tier models like Anthropic's Opus 4.1 but at a tenth of the cost via API.

This will fuel companies like Lovable or Bolt that build entire apps in real-time, making high-performance AI economically viable.

Okay, GPT-5 is live. But is it the game-changer we all expected?

After an initial dive into the specs, the model, and the API, here's my take: The real story isn't one single feature, but the entire package.

🧵 of my thoughts below 👇

GPT 5 is the same level as Opus 4.1 (on SWE bench)

@OpenAI and @AnthropicAI seems to stay near the top together, this is amazing news that top competitors are this close to each other

picture

Claude Code is dancing for me

What does your AI IDE do for you?

https://github.com/vercel/ai/issues/6589
^^ Issue opened

5/5
Thankfully, open source alternatives are starting to catch up in quality.

Maybe it is time to pay more attention to them and invest in solutions that give us more transparency and control.

Would love to hear how others handled this if it happened to you.

Github issue in next comment

4/5
This experience really makes me question how much trust we put in these big platforms.

When the infrastructure is out of your hands, you are always at risk of being blindsided by changes you cannot control.

3/5
I reverted to OpenAI for now, but honestly, I expected more from Google.

If you are releasing a model, revising its version, showcasing it at your keynote, and encouraging everyone to build on it, the least you can do is announce breaking changes and give companies time to adapt.

2/5
I opened an issue and immediately saw a wave of other companies reporting the same thing.

Imagine that your app stops working overnight and you have no idea why. It is a strange feeling when something so critical just fails out of nowhere.

1/5
Yesterday everything worked perfectly in production. Today, Gemini just stopped working.
No code changes. No notice.

I spent hours digging through my own code, convinced I must have broken something, only to finally realize the issue was coming from Google's side.

Google broke my app without any warning.

Full breakdown below 🧵

lies.

how does it look?

Have you ever wanted an open source, ai native presentations?

Fine, I'll build it myself

Are there seriously no open source alternatives to Google Slides????

picture

I love that I'm able to have 3 agents running in parallel on my codebase in cursor

Each has it's own agent mode, planning, executing or critiquing the execution of major features.

For planning I still use dotallio, but the code itself, cursor is hands down the best

5/
For more on the latest and cutting edge AI or my journey as I build @Dotallio feel free to follow me

4/
It seems that OpenAI haven't released any new models, but just released the latest 4o in API access and rebranded it without letting us know.

What do you think about it?

link for the 4o benchmarks - https://artificialanalysis.ai/models/comparisons/gpt-4o-chatgpt-03-25-vs-grok-3

picture

picture

picture

3/
Now let's look at the knowledge cutoff date.

If you access the "latest" 4o model via the api - it shows you "April 2024"

But accessing the 4o model via chat - shows "June 2024" - this is the same as 4.1!

picture

2/
In the benchmark above it shows 4.1 scoring about 66% on GPQA diamond.

I've found one match for GPQA benchmark for the updated 4o 2025-march -
AND IT MATCHES EXACTLY. So the 4.1 model shows the same performance as the march 2025 gpt-4o.

(can see more in the link at the end)

picture

1/
OpenAI claims massive improvements on GPT-4.1 over the GPT-4o - but in the small details you can see in the sub header it says "2024-11-20".

But they've updated their 4o just recently, along with when the image generation came out.

Let's dig in deeper..

picture

OpenAI just lied.

They "Launched" GPT 4.1, even though it appears to be the SAME EXACT MODEL as the recently launched updated 4o on March.

Let me show you the exact details 🧵👇