<?xml version="1.0" encoding="UTF-8" ?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Anurag Jha's Blogs</title>
    <link>https://anuragjha.me/blog</link>
    <description>Blogs by Anurag Jha on full-stack engineering, secure APIs, and AI systems.</description>
    <atom:link href="https://anuragjha.me/rss.xml" rel="self" type="application/rss+xml" />
    <lastBuildDate>Thu, 08 Oct 2026 19:26:39 GMT</lastBuildDate>
    <item>
      <title>Haiku 5.5 vs GPT-6 Luna</title>
      <link>https://anuragjha.me/blog/claude-haiku-5-5-vs-gpt-6-luna-sol/</link>
      <guid>https://anuragjha.me/blog/claude-haiku-5-5-vs-gpt-6-luna-sol/</guid>
      <pubDate>Thu, 08 Oct 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[Below 100,000 input tokens, Claude Haiku 5.5 and GPT-6 Luna have the same listed API rates: $0.10 per million input tokens and $0.50 per million output tokens. Above 100K, Haiku's rates rise to $0.50...]]></description>
    </item>
    <item>
      <title>AI Game Mods Are Getting Easier</title>
      <link>https://anuragjha.me/blog/ai-is-turning-games-into-remixable-systems/</link>
      <guid>https://anuragjha.me/blog/ai-is-turning-games-into-remixable-systems/</guid>
      <pubDate>Wed, 07 Oct 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[People are using coding agents to make game mods and crossovers. In this video, Chaz shows examples and describes three approaches: Pass-through mods add a character or behavior to a game that is alre...]]></description>
    </item>
    <item>
      <title>Chat Works Better When You Can Think Spatially</title>
      <link>https://anuragjha.me/blog/infinite-canvas-llm-chat/</link>
      <guid>https://anuragjha.me/blog/infinite-canvas-llm-chat/</guid>
      <pubDate>Mon, 05 Oct 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[Most chat tools keep everything in one long column. RabbitHoleAI explores a different shape: an infinite canvas where conversations, notes, and drawings can sit next to each other. A chat can branch i...]]></description>
    </item>
    <item>
      <title>Building a Multi-Strategy Options Trading Framework</title>
      <link>https://anuragjha.me/blog/multi-strategy-options-trading-bot/</link>
      <guid>https://anuragjha.me/blog/multi-strategy-options-trading-bot/</guid>
      <pubDate>Mon, 05 Oct 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[This Python project explores how a NIFTY options system can be split into clear parts: strategy signals, execution, market data, and risk controls. Multiple strategies can run alongside one another, w...]]></description>
    </item>
    <item>
      <title>Writing Math on a Canvas, Then Solving It</title>
      <link>https://anuragjha.me/blog/spatial-math-notes/</link>
      <guid>https://anuragjha.me/blog/spatial-math-notes/</guid>
      <pubDate>Mon, 05 Oct 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[Spatial Math Notes is a browser-based take on Apple's Math Notes idea. You write expressions directly on a canvas, and a small TensorFlow.js model recognizes handwritten digits and symbols. Once it se...]]></description>
    </item>
    <item>
      <title>A Marketplace with Separate Buyer and Vendor Flows</title>
      <link>https://anuragjha.me/blog/soko-multi-vendor-ecommerce/</link>
      <guid>https://anuragjha.me/blog/soko-multi-vendor-ecommerce/</guid>
      <pubDate>Mon, 05 Oct 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[Soko is a multi-vendor marketplace built with the MERN stack. It separates the needs of shoppers, vendors, and administrators, with role-based dashboards and APIs for the different parts of the market...]]></description>
    </item>
    <item>
      <title>Spring Physics for UI Motion</title>
      <link>https://anuragjha.me/blog/motionspring/</link>
      <guid>https://anuragjha.me/blog/motionspring/</guid>
      <pubDate>Mon, 05 Oct 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[CSS easing is great for ordinary transitions, but sometimes I want motion that responds like a spring: it can overshoot, settle, and feel connected to the distance it needs to travel. MotionSpring is...]]></description>
    </item>
    <item>
      <title>A Neural Network Small Enough to Understand</title>
      <link>https://anuragjha.me/blog/neural-network-from-scratch/</link>
      <guid>https://anuragjha.me/blog/neural-network-from-scratch/</guid>
      <pubDate>Mon, 05 Oct 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[Most neural network tutorials jump quickly to frameworks and hide the math behind an API. This project goes the other way: a tiny two-layer network in NumPy, trained on XOR. The forward pass turns inp...]]></description>
    </item>
    <item>
      <title>Batching LLM Calls</title>
      <link>https://anuragjha.me/blog/batching-llm-calls/</link>
      <guid>https://anuragjha.me/blog/batching-llm-calls/</guid>
      <pubDate>Mon, 05 Oct 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[If several small questions can be answered independently, sending each one as its own request adds overhead. This little Python experiment puts multiple queries into one prompt, labels each with an ID...]]></description>
    </item>
    <item>
      <title>GPTCoderRalph: Connecting ChatGPT to Local Development</title>
      <link>https://anuragjha.me/blog/gptcoderralph/</link>
      <guid>https://anuragjha.me/blog/gptcoderralph/</guid>
      <pubDate>Mon, 05 Oct 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[GPTCoderRalph explored a practical gap in browser-based coding: ChatGPT can discuss a code change, but the code lives in a local development environment. The extension was meant to bridge those two pl...]]></description>
    </item>
    <item>
      <title>Making Transparent PNGs from Generated Images</title>
      <link>https://anuragjha.me/blog/nanobanana-png/</link>
      <guid>https://anuragjha.me/blog/nanobanana-png/</guid>
      <pubDate>Mon, 05 Oct 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[Image generators are good at making a subject, but they do not always return the clean transparent background a design workflow needs. NanoBanana PNG tries a simple workaround: ask Gemini for the subj...]]></description>
    </item>
    <item>
      <title>A Camera App with a More Direct Image Pipeline</title>
      <link>https://anuragjha.me/blog/camera2api-custom-tone-mapping/</link>
      <guid>https://anuragjha.me/blog/camera2api-custom-tone-mapping/</guid>
      <pubDate>Mon, 05 Oct 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[Camera2API is an Android camera experiment built around the controls I wanted from a camera app: manual settings and a less processed look. It includes log and quadratic Bézier tone curves, plus toggl...]]></description>
    </item>
    <item>
      <title>A Search and Research Assistant in the Terminal</title>
      <link>https://anuragjha.me/blog/perplexity-clone-cli/</link>
      <guid>https://anuragjha.me/blog/perplexity-clone-cli/</guid>
      <pubDate>Mon, 05 Oct 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[This CLI takes a query, searches the web, extracts text from a result, and asks Gemini to turn that material into a structured Markdown answer. It uses Trafilatura to pull readable page content out of...]]></description>
    </item>
    <item>
      <title>Turning a Video Spec into a Captioned Clip</title>
      <link>https://anuragjha.me/blog/text-to-caption-video-generator/</link>
      <guid>https://anuragjha.me/blog/text-to-caption-video-generator/</guid>
      <pubDate>Mon, 05 Oct 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[This project explored generating a video from a compact specification: describe the scenes, provide the text, and assemble the result with captions and multiple audio tracks. MoviePy handles the media...]]></description>
    </item>
    <item>
      <title>Quick Image Corrections from the Command Line</title>
      <link>https://anuragjha.me/blog/tiny-image-correction-cli/</link>
      <guid>https://anuragjha.me/blog/tiny-image-correction-cli/</guid>
      <pubDate>Mon, 05 Oct 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[Sometimes an image only needs a small adjustment. This tiny Python and PIL command-line tool focuses on brightness, saturation, and contrast, keeping those edits quick to apply without opening a full...]]></description>
    </item>
    <item>
      <title>What People Are Doing with OpenAI Dots</title>
      <link>https://anuragjha.me/blog/openai-dots/</link>
      <guid>https://anuragjha.me/blog/openai-dots/</guid>
      <pubDate>Thu, 01 Oct 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[OpenAI dropped "Dots" at DevDay alongside GPT-6.1 Sol. Instead of a chat box where you type a prompt and wait, a Dot is an always-on agent. You assign it a job, close your laptop, and it runs in the b...]]></description>
    </item>
    <item>
      <title>Gemini 4 Argon</title>
      <link>https://anuragjha.me/blog/gemini-4-argon/</link>
      <guid>https://anuragjha.me/blog/gemini-4-argon/</guid>
      <pubDate>Thu, 01 Oct 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[Anthropic dropped Sonnet 5.5 on Sunday. OpenAI dropped GPT-6.1 Sol and Dots on Tuesday. Now Google dropped Gemini 4 Argon. Four frontier drops in eight days. Model shelf life is officially cooked, lol...]]></description>
    </item>
    <item>
      <title>GPT-6.1 Sol</title>
      <link>https://anuragjha.me/blog/gpt-6-1-sol/</link>
      <guid>https://anuragjha.me/blog/gpt-6-1-sol/</guid>
      <pubDate>Wed, 30 Sep 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[Anthropic dropped Sonnet 5.5 on Sunday. Exactly 24 hours later, OpenAI dropped GPT-6.1 Sol. Model shelf life is down to 24 hours now, lol. OpenAI GPT-6.1 Sol (Sept 29) List price stays at $2/M in and...]]></description>
    </item>
    <item>
      <title>Is the Dunning-Kruger Effect Real?</title>
      <link>https://anuragjha.me/blog/is-dunning-kruger-real/</link>
      <guid>https://anuragjha.me/blog/is-dunning-kruger-real/</guid>
      <pubDate>Tue, 29 Sep 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[Partly right. The critique you read is real and pretty strong, but "Dunning-Kruger isn't real" overstates it a bit. The famous 1999 study sorted people into quartiles by test score and then plotted ho...]]></description>
    </item>
    <item>
      <title>Claude Sonnet 5.5</title>
      <link>https://anuragjha.me/blog/claude-sonnet-5-5/</link>
      <guid>https://anuragjha.me/blog/claude-sonnet-5-5/</guid>
      <pubDate>Mon, 28 Sep 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[Anthropic dropped Claude Sonnet 5.5 six days after Opus 5.5. Anthropic Claude Sonnet 5.5 (Sept 28) List price stays at $2/M in and $10/M out. It's 30% faster and uses fewer tool calls to finish tasks....]]></description>
    </item>
    <item>
      <title>Claude Opus 5.5 and GPT-6 Sol</title>
      <link>https://anuragjha.me/blog/claude-opus-5-5-and-gpt-6-sol/</link>
      <guid>https://anuragjha.me/blog/claude-opus-5-5-and-gpt-6-sol/</guid>
      <pubDate>Wed, 23 Sep 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[Three weeks after the Astra and Fable drops, Anthropic, OpenAI, and Xiaomi dumped new models on the exact same Tuesday. Anthropic dropped Claude Opus 5.5. OpenAI dropped GPT-6 Sol and GPT-6 Luna. Xiao...]]></description>
    </item>
    <item>
      <title>Building a Harness with Jev</title>
      <link>https://anuragjha.me/blog/making-a-harness-with-jev/</link>
      <guid>https://anuragjha.me/blog/making-a-harness-with-jev/</guid>
      <pubDate>Fri, 18 Sep 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[Building a Harness with Jev Jev is a model released by TypeSafe AI. The company reports up to 200x faster inference and 400x lower cost than comparable LLMs on classification tasks, trained using Rein...]]></description>
    </item>
    <item>
      <title>The OpenAI Navier Stokes Codex Controversy</title>
      <link>https://anuragjha.me/blog/openai-gpt-7-navier-stokes/</link>
      <guid>https://anuragjha.me/blog/openai-gpt-7-navier-stokes/</guid>
      <pubDate>Wed, 09 Sep 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[OpenAI just dropped a paper claiming a full solution to the Navier-Stokes existence and smoothness problem. It is one of the seven Clay Millennium Prize problems, open for 92 years since Jean Leray's...]]></description>
    </item>
    <item>
      <title>Anthropic Researcher Jacob Coxon Resigned</title>
      <link>https://anuragjha.me/blog/anthropic-researcher-jacob-coxon-resigned/</link>
      <guid>https://anuragjha.me/blog/anthropic-researcher-jacob-coxon-resigned/</guid>
      <pubDate>Wed, 09 Sep 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[Jacob Coxon spent the last three years doing pretraining research at both OpenAI and Anthropic. Today he resigned. His post on X hit 18M views in five hours. His message: neither lab is acting respons...]]></description>
    </item>
    <item>
      <title>What is GPT-6 Provider Adapter</title>
      <link>https://anuragjha.me/blog/gpt-6-astra-provider-adapter/</link>
      <guid>https://anuragjha.me/blog/gpt-6-astra-provider-adapter/</guid>
      <pubDate>Sat, 05 Sep 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[In my last post I wrote Astra hit 99.9% on ARC-AGI-3 and called the benchmark dead. But the leaderboard shows two numbers: 62.7% and 99.9%. The 37% gap is the harness. When ARC-AGI-3 launched 6 months...]]></description>
    </item>
    <item>
      <title>GPT-6 Astra Gemini 3.8 Muse Spark 1.3 All at Once</title>
      <link>https://anuragjha.me/blog/gpt-6-astra-gemini-3-8-muse-spark-1-3-all-at-once/</link>
      <guid>https://anuragjha.me/blog/gpt-6-astra-gemini-3-8-muse-spark-1-3-all-at-once/</guid>
      <pubDate>Fri, 04 Sep 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[In my last post I wrote OpenAI was quiet and might drop something soon against Anthropic. Well, OpenAI didn't just drop something. Everyone dropped. In literally 48 hours: Gemini 3.8 Flash dropped, Me...]]></description>
    </item>
    <item>
      <title>Claude Fable 5.1</title>
      <link>https://anuragjha.me/blog/claude-fable-5-1/</link>
      <guid>https://anuragjha.me/blog/claude-fable-5-1/</guid>
      <pubDate>Wed, 02 Sep 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[The latest OpenAI model GPT-5.6 went public on July 9, 2026. Meanwhile, Anthropic has dropped two frontier models: Opus 5 and now Fable 5.1. That's wild, OpenAI might drop a new model soon. Anthropic...]]></description>
    </item>
    <item>
      <title>How to Make Real Money with GTA 6</title>
      <link>https://anuragjha.me/blog/how-to-make-money-with-gta-6/</link>
      <guid>https://anuragjha.me/blog/how-to-make-money-with-gta-6/</guid>
      <pubDate>Sun, 30 Aug 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[GTA 6 is an economy. Players spend real cash on in-game progress, millions watch streams, and attention scales to billions of eyeballs. Building tools or content around it makes real money. Interactiv...]]></description>
    </item>
    <item>
      <title>Million Product Companies</title>
      <link>https://anuragjha.me/blog/rise-of-million-product-companies/</link>
      <guid>https://anuragjha.me/blog/rise-of-million-product-companies/</guid>
      <pubDate>Sat, 28 Mar 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[It has never been easier to build a product. Coding is no longer the bottleneck. The bottleneck is figuring out what people actually want (which has always been hard, but now it's the only hard part)....]]></description>
    </item>
    <item>
      <title>how to fix ai slop</title>
      <link>https://anuragjha.me/blog/how-to-fix-ai-slop/</link>
      <guid>https://anuragjha.me/blog/how-to-fix-ai-slop/</guid>
      <pubDate>Tue, 10 Mar 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[Code used to be the hard part of software (was it?). Now it isn't. The hard part is building systems that don't break (or rather break gracefully). With AI, writing code has become a byproduct of the...]]></description>
    </item>
    <item>
      <title>which language takes least ai tokens ?</title>
      <link>https://anuragjha.me/blog/language-ai-tokens/</link>
      <guid>https://anuragjha.me/blog/language-ai-tokens/</guid>
      <pubDate>Sun, 15 Feb 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[When prompting AI models, token count directly affects both cost and latency. This benchmark compares how efficiently different languages express similar logic. Ranking across available tests (Stats F...]]></description>
    </item>
    <item>
      <title>Journey</title>
      <link>https://anuragjha.me/blog/journey/</link>
      <guid>https://anuragjha.me/blog/journey/</guid>
      <pubDate>Thu, 22 Jan 2026 00:00:00 GMT</pubDate>
      <description><![CDATA[Started coding around 2014–15; I don't remember exactly. Why? Games! My first language was MSW Logo, then QBASIC also VBA. Then C++; here’s my oldest online coding profile. Then Java, followed by SQL...]]></description>
    </item>
  </channel>
</rss>