ChatGPT Just Got Instantly Better
Plus, 🎥 How to Catch Brand Mentions Hiding Inside Videos, Google and Alibaba are betting on faster AI models, and more.
Hola Decoder😎
If someone forwarded this to you and you want to Decode the power of AI and be limitless, then subscribe now and Join Decode alongside 30k+ code-breakers untangling AI.
🤖 OpenAI Releases GPT-5.3 Instant with Focus on Tone, Accuracy, and Fewer Refusals
OpenAI's latest update to ChatGPT's default model isn't about new features. It's about making the existing experience less frustrating, with better answers, fewer dead ends, and a more natural tone.
Fewer Unnecessary Refusals and Disclaimers: GPT-5.2 was often overly cautious, refusing safe questions or adding preachy caveats. 5.3 cuts that back significantly. When a useful answer is appropriate, it gives one directly without the moralizing warmup.
Smarter Web Search Answers: The model now blends web results with its own knowledge instead of dumping loosely connected links. It picks up on subtext better and puts the most relevant information upfront rather than burying it.
Measurably Fewer Hallucinations: On high-stakes topics like medicine, law, and finance, hallucination rates dropped 26.8% with web access and 19.7% without. On user-flagged errors, the drop was 22.5% and 9.6%, respectively.
What This Means
This is OpenAI responding directly to user complaints about tone and usability. For anyone building on or competing with ChatGPT, the bar for conversational AI just moved. The focus on reducing refusals and improving synthesis over raw capability signals is where the product race is heading.
Together with Modash
Your Affiliate Program Might Be Profitable on Paper. That Doesn’t Mean It Works.
You launch an affiliate program. A few creators join. Some sales come in. Then growth plateaus.
Not because affiliate doesn’t work, but because the math, structure, and incentives were never designed to compound.
Modash just released the complete A-Z guide to building an affiliate program that actually holds up under scale.
Inside, you’ll discover:
- Why commission rates mean nothing without understanding margin compression
- How most tier systems accidentally demotivate top affiliates
- How code leakage and cookie gaps quietly distort attribution
This isn’t a surface-level playbook. It’s built from working with thousands of creators and real-world programs.
If you’re building affiliate as a real growth channel in 2026?
🎥 How to Catch Brand Mentions Hiding Inside Videos with Syncly Social
Most listening tools track text. But in 2026, brand impact happens in spoken words, product placements, and untagged video moments.
Here’s how to use Syncly Social to stay ahead:
1️⃣ Set up your brand monitoring
Add your brand name, products, and competitor terms. Syncly scans TikTok, Instagram, and YouTube at the video level, not just captions.
2️⃣ Uncover spoken and untagged mentions
Find creators who mention or show your product without tagging you. These are the hidden growth or risk signals.
3️⃣ Track emerging narratives early
Spot negative sentiment or viral spikes before they explode in comments or mainstream media.
4️⃣ Identify creators already converting
Use influencer discovery to find people who are already moving your audience naturally.
Instead of reacting late, you see brand momentum the moment it starts. You can get started here for free.
⚡ Google and Alibaba Are Betting on Cheaper, Faster, Smaller AI Models
The AI race is shifting. Instead of chasing the smartest model, Google and Alibaba both dropped new releases optimized for speed, cost, and running on smaller hardware. The focus is now on making AI feel like infrastructure.
Google’s Flash-Lite Is Built for Scale
Gemini 3.1 Flash-Lite is priced at $0.25 per million input tokens, a fraction of competitors. It’s 2.5x faster to the first token than Gemini 2.5 Flash with adjustable reasoning levels per task. Built for companies making millions of API calls daily.
Alibaba’s Qwen 3.5 Runs on Your Phone
Four models ranging from 0.8B to 9B parameters, all able to run locally with no cloud needed. The 9B version uses Scaled Reinforcement Learning to compete with models 5 to 10x its size on accuracy and reasoning.
What this means
If you’re building products or workflows on top of AI, cost per call and latency matter more than ever. Google is targeting enterprise volume. Alibaba is targeting on-device independence. Both are signals that the “good enough and everywhere” era of AI is arriving fast.
Together with Norton Neo
Trust-First AI Built Directly Into Your Browser
Most AI tools sit on top of your browser. Norton Neo builds AI directly into it.
Norton Neo is the world’s first AI-native browser designed for speed, clarity, and privacy from the ground up.
You browse, search, stream, shop, and work as usual, but with AI that organizes your tabs, adapts to how you work, and helps you move without constant prompting.
- Chats and browsing history remain local, ensuring your data stays private.
- You choose what Neo remembers, or turn memory off anytime.
- Norton Web Shield blocks malicious sites and phishing before you click.
No extra extensions stacking up. No AI tools floating in separate tabs. Just a browser built for how people actually use the web today.
If you want early access to what’s coming next, join the Norton Neo Discord and connect directly with the team building it.
🔨Tools you cannot miss:
🤖 Viktor - An AI agent that automates tasks, writes code, connects to 3,000 tools, and executes real work instead of just answering questions.
✂️ ClipFinder - Instantly turn long-form videos and streams into viral-ready short clips in seconds.
🎧 PodShrink - Condense hours of podcasts into quick, easy-to-read summaries so you get the key insights in minutes.
⌨️ Clico - An AI assistant that works directly where you type on any website, eliminating copy-paste workflows with a single shortcut.
🧩 Prompt of the Day
Loyalty Program Integration Framework for a New Product
Turn a new product launch into a repeat-purchase engine by embedding it directly into your loyalty ecosystem.
Paste the prompt: Drop this into ChatGPT, then fill in your product name, reward structure, and promotional channels.
Prompt to paste
Create a loyalty program integration plan for [Insert Product]. Include:
- Points Earned Per Purchase: Define how many loyalty points customers earn when buying the product and whether there are bonus incentives (e.g., first purchase bonus, double points periods, bundle bonuses).
- Exclusive Rewards: List loyalty benefits tied to the product, such as early access, member-only bundles, special editions, or redemption perks.
- Promotion Plan: Outline how the loyalty integration will be promoted across email, website/app, social media, SMS, and in-store messaging if relevant.
- Use Case: Focus on how this integration increases repeat purchases, accelerates product adoption, and strengthens customer engagement.
🤳AI Nugget of the Day
Thanks for Decoding with us🥳
Your feedback is the key to our code! Help us elevate your Decode experience by hitting reply and sharing your input on our content and style.
Keep deciphering the AI enigma, and we'll be back with more coded mysteries unraveled just for you!