<h1 id="2025-04-16">2025-04-16<a aria-hidden="true" class="anchor-heading icon-link" href="#2025-04-16"></a></h1>
<h1 id="o3-and-o4-mini-release">o3 and o4-mini release<a aria-hidden="true" class="anchor-heading icon-link" href="#o3-and-o4-mini-release"></a></h1>
<p><a class="color-tag" style="--tag-color: #fff4f2;" href="/dendron-wiki/notes/jcbadkzec97t7sa62s5q49c">#AI</a></p>
<p><a href="https://www.youtube.com/watch?v=sq8GBPUb3rk&#x26;t=823s">youtube</a></p>
<p>Today OpenAI released its latest reasoning models. LOL - these model reach performance equivalent to the top 200 programmers in Codeforces.. this is super crazy. The presenters say that the core idea behind the technology is next token prediction.. And this works like magic - very impressive to be honest.</p>
<p>For the first time the reasoning models can use tools: code in Python, search the web, the terminal, analyzing uploaded files and other data with Python, reasoning deeply about visual inputs, and even generating images. So it is <strong>multimodal</strong>.</p>
<p>This reasoning models can browse your codebase and fix your code. It shows it logic, explanation and verifies the solution.</p>
<p>o3 is cheaper than o1, both training and inference. </p>
<p>They released codeX CLI tool that has this extra reasoning</p>

2025-04-16


Welcome to my Knowledge Base! Here I write about my perception of life, document exciting things I've learned, debate (with myself) on controversial topics. If you know me you will not be surprised to find out that I write mostly about engineering and maths. Other topics I'm interested in are economics, politics, business, chess and poker.