In partnership with
๐ฑ
Inception just dropped Mercury 3, and it's fast. Really fast.
This isn't your typical AI model that spits out words one at a time it generates entire chunks of text at once using diffusion, the same tech that powers image generators.
But wait, there's more:
A major security disaster hit Python and Linux users when an Indian certificate authority screwed up their encryption keys
OpenAI is now showing ChatGPT ads in 89 countries because apparently 100,000 advertisers want in
Iโm Alex. Welcome to L8R by Innov8.
Letโs dive deep.
In todayโs post:
Inception's Mercury 3 rewrites the speed game with diffusion
Indian CA breach exposes Python and Linux users to forgery attacks
OpenAI pushes ChatGPT ads into 26 more countries
Mercury 3: The 1,000 Tokens-Per-Second AI
AI speed limits just shattered.
Inception dropped Mercury 3 on October 19, 2026, and it writes text in parallel instead of word-by-word.
This weird diffusion trick makes it blazing fast for complex AI agents.
The details:
Mercury 3 hits speeds over 1,000 tokens per second on NVIDIA GPUs by drafting whole blocks of text at once.
The model handles a massive 1 million token context window and works perfectly with existing OpenAI setups.
Inception claims it beats GPT-6 Luna and Claude Haiku 5.5 on most tests, especially in coding and following instructions.
You can grab standard access for $0.20 per million input tokens, but they are running a 50 percent off launch promo right now.
Older versions of Mercury are shutting down on November 15, 2026.
Why it matters:
Normal AI models write left to right.
That takes way too long when you chain multiple AI agents together.
Mercury 3 proves you can use image-style diffusion tricks for text at a massive scale.
If this works as well as they say, your AI assistants are about to get instant reaction times.
๐ก L8R's Take:
I love the crazy speed, but I hate the secrecy.
Inception refuses to release the open weights for independent testing.
Until someone else proves it actually beats Claude Haiku 5.5, treat their internal charts like a slick marketing pitch.
From Our Partner
Why Most AI Tools Fail at B2B Customer Service
B2B customer issues rarely resolve in a single interaction. They move across teams, systems, and account history that most AI tools were never designed to track.
A new briefing paper from Harvard Business Review Analytic Services, sponsored by Front, looks at why AI built for high-volume consumer support creates new coordination gaps when it meets the multi-team reality of B2B service, and what leaders should ask before investing in their next AI tool.
Get the free briefing paper to see the three questions that separate tools built for your environment from tools that will create more work.
Update Python Now: A Massive Security Flaw Just Broke Trust
A massive security disaster just hit the web.
Indian certificate authority eMudhra accidentally generated incredibly weak private keys, letting anyone easily forge trusted websites.
If you use Python or Linux, your systems might trust these fake sites right now.
The details:
Eight intermediate certificates from eMudhra had keys so small (one had a value of just 4) that anyone could guess them instantly.
These bad keys were made back in April 2025 due to a script error, meaning the web was secretly vulnerable for over a year.
Major players like Apple, Google, and Microsoft already blocked the bad certificates on October 20, 2026.
Python users must update to uv 0.13.4, pip 26.2.2, and certifi 2026.10.19 to stay safe.
Linux users need to update their ca-certificates packages immediately to stop hackers from intercepting their traffic.
Why it matters:
The internet runs on trust.
When your browser sees a padlock, it trusts the certificate authority behind it. eMudhra broke that trust chain completely.
Because these keys were guessable for over a year, bad actors had plenty of time to spy on encrypted traffic without anyone noticing.
๐ก L8R's Take:
This is an embarrassing mistake for a security company.
You cannot mess up a basic math script and leave the whole internet exposed for 18 months.
Go update your Python packages and Linux servers right now, and never blindly trust a single authority again.
From Our Partners
Leave Granola and get up to 12 months free of Wispr Flow Notetaker + Dictation
If you have paid time left on an individual Granola plan, we'll match it with a Wispr Flow subscription that includes Notetaker and dictation, and add bonus time, up to 12 months total. Sign in or create a Wispr account and submit proof of your plan to check eligibility.
ChatGPT Ads Hit 89 Countries
Free ChatGPT is about to get a lot more crowded with ads.
OpenAI just pushed its advertising platform into 26 more countries, bringing the total to 89 markets worldwide.
The AI giant is turning its massive user base into a serious cash cow ahead of a rumored 2027 IPO.
The details:
Over 100,000 advertisers are now using the platform, a huge jump from just 20,000 in August 2026.
Ads will only show up on the Free and Go plans, keeping Plus, Pro, and Enterprise tiers totally ad-free.
The new expansion adds 14 Latin American countries and covers spots across Europe, Africa, and Asia.
OpenAI's ad business is already hitting a massive $1 billion revenue run rate.
Rival company Anthropic is taking the opposite path and promises to keep Claude ad-free.
Why it matters:
Running AI models costs a huge amount of money.
OpenAI is following the classic tech playbook by keeping premium users happy while making money off free users.
They are even testing interactive 'Sponsored Agents' instead of just boring banners.
This is all about proving they can make real profit before they go public.
๐ก L8R's Take:
Nobody likes ads, but this was always going to happen.
You cannot give away world-class AI for free without someone paying the bill.
If you really hate seeing ads, you will just have to pay for a Plus plan or switch to Claude.
๐ Quick L8R Summary
Mercury 3: Inception dropped a diffusion-based language model that generates text in parallel blocks instead of word-by-word claims it's way faster than traditional models.
eMudhra breach: Indian CA company's weak crypto keys got cracked, putting Python and Linux users at risk major supply chain security mess.
ChatGPT ads everywhere: OpenAI just pushed its advertising platform to 26 more countries, now reaching 89 markets with over 100,000 advertisers.
๐ฉ Innathe L8R engane undarunnu ๐?
We read every reply - just reply to this email and let us know how we can improve !
Appo adutha L8R il kanaam bieโฆ๐