<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Nikhil Kumar on Medium]]></title>
        <description><![CDATA[Stories by Nikhil Kumar on Medium]]></description>
        <link>https://medium.com/@uk.nikhil2004?source=rss-796ad2804139------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*f-poqu7sxnd-2Xig</url>
            <title>Stories by Nikhil Kumar on Medium</title>
            <link>https://medium.com/@uk.nikhil2004?source=rss-796ad2804139------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 12:50:04 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@uk.nikhil2004/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Initial Thoughts on GPT-5(Crazy week in the community)]]></title>
            <link>https://medium.com/@uk.nikhil2004/initial-thoughts-on-gpt-5-crazy-week-in-the-community-d17e27a10490?source=rss-796ad2804139------2</link>
            <guid isPermaLink="false">https://medium.com/p/d17e27a10490</guid>
            <dc:creator><![CDATA[Nikhil Kumar]]></dc:creator>
            <pubDate>Sun, 10 Aug 2025 04:00:13 GMT</pubDate>
            <atom:updated>2025-08-10T04:00:13.785Z</atom:updated>
            <content:encoded><![CDATA[<p>Hey Guyss!! This is my first Medium article, and honestly, it felt like the right time or maybe the right week to start writing about AI. I’ve been lurking in the community for a while, alerting my juniors and friends with the tech checks, lol. Alright, time to be serious for at least one time so i picked up Medium, and here we are. So Grab your coffee as we get into what just went down 🚀</p><p>Okay as you have seen the what the title is about and everything that happened this week but right before we deep dive into the GPT -5 which OpenAI team released on 7th of August(07.08.2025) but before that something major happened OpenAI released 2 Open Source models on 5th of August (05.08.2025). Yeh you heard me right Sama and team actually gave us two open source models .</p><p>Wait what is OpenSource??????? People who feel like this is a jargon ill make it very simple. Lets say (for ex) XOXO is releasing a model as a OpenSource that means we can download them, modify them and use them commercially without asking for permission. Wild right?</p><p>Okay here are the two models Meet gpt-oss-120b and gpt-oss-20b(<a href="https://openai.com/open-models/">Open models by OpenAI | OpenAI</a>). We’re looking at a 117 billion parameter beast and its smaller 21 billion parameter sibling, both designed specifically for reasoning tasks and agentic workflows. I mean, this is the same company that kept GPT-3 locked behind an API for years and made us all beg for access. Remember when getting into the GPT-4 beta felt like winning the lottery? And now they’re just… giving stuff away? A very Strategic move.</p><p>The community went absolutely wild. Twitter was buzzing, Reddit was losing its mind, and suddenly everyone was spinning up these models and sharing results. This felt like watching a teaser showcase before the main trailer lol.</p><p>Then came the main picture GPT-5!!!! phew okay just like everyone i was so hyped and tuned in to watch the live stream IST 10.30 PM props to (<a href="https://www.linkedin.com/in/elaine-ya-le/">https://www.linkedin.com/in/elaine-ya-le/</a>) she really cooked the show out there with Sama and team. Back to business as they shared and showed the demo of whats the maximum potential of their most best Frontier Intelligence can do i felt like a good competition for Claude’s opus 4.1 then i saw this 😵‍💫 (Attaching the img down)</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/632/1*LmhhqlQGbrtpDkKCvtJh1g.png" /></figure><p>This was just top tier like dawg look at the SWE Bench verification to put it in context lets have a look at the Opus 4.1 from Anthropic</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*X7aX0NIVQ1MUzf_XVtIJIw.png" /></figure><p>That’s exactly why i mentioned a healthy competiton they are going toe to toe. One thing when i personally started using the free credits of GPT-5 in Cursor Ai terminal i felt like it has better understanding and gives a even better and improved ui and has the initial capacity for a good frontend struture (Better than 4o imo) but the issue with GPT-5 in a nutshell is that unless you pay for model switching &amp; know to use GPT-5 Thinking or Pro, when you ask “GPT-5” you sometimes get the best available AI &amp; sometimes get one of the worst AIs available and it might even switch within a single conversation(So many cons tbh)</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*UXvO7JEaf5pZwWll" /><figcaption>GPT-5 Thinking (Just got the award for worst model on IQ test)</figcaption></figure><p>Context window(a limit token to understand the contexts and remembering within a single chat) of GPT-5 is around 400k(272k for i/p and 128k for o/p) tokens whereas in this case Gemini 2.5 flash has around 1 millions tokens (Google leads the chart here). Lets not forget that Google are about to release Gemini 3.0 lol.</p><p>The biggest mistake they did was annoucing GPT-5 as the frontier and unified solution cuz they deleted all the other 8 models including the fan favourite GPT-4o 🕊️🪦.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/900/0*vkarYBKq3zUDpa6M" /><figcaption>People’s Opinion on GPT-5</figcaption></figure><p>At this point, I feel like the benchmark comparisons have become saturated. Moving forward, the context window the model’s ability to remember and understand a long conversation will be the most crucial factor. The demo of GPT-5 was good but all depends upon the live use cases which can increase the industry sales they both go hand in hand.</p><p>Despite Flaws even if GPT-5 is viewed comparatively critically in the AI community, I am of the opinion that it will not detract from the success of ChatGPT. On the contrary for 95% of everyday users, GPT-5 represents a significant leap in performance. If only because people will experience reasoning via router for the first time. Overall, GPT-5 is therefore likely to be a great global success.</p><p>Now we wait for Gemini 3….lol see y’all in my next blog it was fun ranting things which were running on my mind for the past few days ill be back with another blog soon and also if you love to share your opinion or advise me how to blog pls do lmk constructive criticism works well for me 🫶.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d17e27a10490" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>