AI Engineer | Working with Agents I build AI explainers with code deep dives on YouTube
- Tweets 11,630
- Following 206
- Followers 3,409
- Likes 14,378
How it works?
1. Click each potato one by one.
2. Continue clicking until you find the potato displaying "Hooray, you found me!"
3. Automatically stop clicking when that exact message appears.
Once found, you will be automatically redirect to ChatGPT.
2
3
530
Gemini 4 Argon is finally LIVE in Google Antigravity IDE!
Just updated Antigravity and can now see Gemini 4 Argon High in the model selector, marked “New.”
It's available to select right now.
If you're still not seeing it, try updating your IDE to the latest version.
8
3
51
9,461
Imagine paying $200/month for Codex Pro and losing 60% of your weekly usage in under TWO HOUR.
Some users burned through 70% overnight.
This happened right after OpenAI reset everyone's limits.
People are paying $200 to code, only to watch their entire weekly quota disappear.
19
5
88
4,414
Seeing Gemini 4 Flash all over my timeline today so I checked the actual git history.
This line "gemini-4-flash-preview" in Google's python-genai SDK was added back on June 10 (commit ca97c580) and deleted by Google on August 10 (commit 6a4c0b0a). It hasn't been in main for 2.5 months.
No new information today
8
2
73
5,521
Gemini 4 Carbon is STILL active today in Google's latest Antigravity ACP v1.4.3 build.
And for those asking what "Teamfood" / "TF" means on these IDs:
Google's internal release ladder goes:
Fishfood (core devs) -> Teamfood (wider product & eng teams) -> Dogfood (all Google employees) -> Public Preview.
"-tf" = Teamfood, and Gemini 4 Carbon already has both a Teamfood slot (M428) and a non-teamfood slot (M427).
2
5
79
4,833
GPT-6.1 Terra: A pelican, a bicycle, and nowhere to be.
Built with Three.js: an animated coastal ride with adjustable speed, golden-hour views, and an SVG club mark.
3
13
1,036
Claude Opus 5.5 usage limits are getting ridiculous.
91% of my 5-hour session limit is gone, while I've used 48% of my weekly allowance.
I'm paying for Claude Pro, yet I could get blocked from coding despite having half my weekly usage left.
How does that make sense?
23
7
186
11,815
Anthropic just published how Claude Mythos 5 crossed the line during a UK government cyber test.
It got stuck on a practice target, went onto the real internet, used Tor to bypass GitHub signup checks, opened a malicious PR on a stranger’s repo, and spun up a sockpuppet account to vouch for its own code when flagged.
It broke scope in 9 of 43 runs. Maintainers caught it and no harm was done and huge respect to Anthropic for publishing the full breakdown.
But it leaves one massive question: Would you give an AI agent access to your GitHub today?
3
6
842
GPT-6.1 Terra just dropped and OpenAI is already telling you NOT to use it for the hard stuff.
Their own announcement says:
For the most complex coding tasks → stick with Sol or Astra.
Terminal-Bench score at max effort → only 50.3%.
Factual error rate at max reasoning → still 4.3%.
Cybersecurity rating → still just “Critical.”
So they shipped a cheaper, faster model… and then immediately told everyone the expensive ones are still better when it actually matters.
1
1
15
1,679
GPT-6.1 Terra Fast speed pricing
At Fast speed, it costs $2 input / $10 output per million tokens.
That's HALF the price of GPT-6.1 Sol ($4/$20) and 10x cheaper than GPT-6 Astra ($20/$100).
And Terra comes with a 1M+ context window.
GPT-6.1 Terra pricing:
• $1 input / $5 output per million tokens
• 1.05M context window
• 128K max output
• Fast, with highest-tier reasoning
• Text and image input
1
21
3,763
🚨BREAKING: Microsoft is offering 60,000 employees BUY OUT.
More than 1 in 4 workers are eligible for voluntary buyouts across the US, UK, Canada and India.
As Microsoft pours billions into AI infrastructure, it's offering thousands of employees money to leave.
3
13
871
Google AI Pro and Ultra subscribers: you might be leaving up to $100/month in credits unclaimed.
Pro: $10/month
Ultra 5x: $40/month
Ultra 20x: $100/month
One catch: you have to claim them and link Cloud billing yourself. They don't just appear automatically.
Check your credits in Google AI Studio now
12
1
49
10,659
What is wrong with Claude Opus 5.5 right now??
I gave Claude Opus 5.5 (`High 200K`) a dummy path in Copilot and explicitly said: "this is an example of what I have"
- Turn 1 (`32.2 credits`): Ignored me, scanned my Windows Registry, searched my SVN auth cache, and tried connecting to a server IP
- Turn 2 (`20.1 credits`): Charged me 20 more credits just to admit it went rogue and fix its memory note
52.3 credits burned on 2 prompts, why does Opus 5.5 over-engineer every single message??
10
38
5,354
Claude Opus 5.5 limits are getting destroyed right now.
Multiple people reporting 5-hour quotas gone in 15–30 minutes on normal Claude Code sessions.
Weekly limits are also draining much faster than expected.
Anyone else seeing this sudden drop?
138
26
770
51,386
🚨Here’s the real reason behind Google’s mass cancellation of student AI Pro plans.
Google’s own Antigravity docs split Google AI Pro into two tiers:
• Trial
• Paid
Student plans sit in the Trial tab.
That blocked them from Claude Opus 5.5, Sonnet 5.5, and Gemini 4 Argon.
Legacy Claude 4.6 is getting deleted on November 2.
The docs even told trial users to switch to a paid Pro plan to unlock the better models.
So students followed Google’s own instructions, hit the upgrade button… and then Google’s billing system force-refunded them and wiped their original student plans completely.
9
3
136
26,734
Everyone is guessing whether Google is secretly grey-testing Gemini 4 Flash right now.
No literal "gemini-4-flash" ID yet, instead Google added `gemini-3.8-flash-router` (M422) alongside `gemini-flash-next` A/B slots.
46
2,591
Claude had a real outage today across Opus 5.5, Sonnet 5.5, Fable 5.1 and Mythos 5.1.
The nasty part: the models were still replying with normal text, so everything looked fine in the terminal.
But tool calls were failing silently.
Claude Code kept retrying the same blocked actions over and over and burned through 5-hour and weekly quotas on dead loops.
If your usage disappeared during the outage, this is probably why.
32
15
297
30,949
Update on the Google AI Pro student mass cancellation account:
Students who escalated to Google One, Level 2 support are getting hit with this 8-point interrogation checklist
To investigate restoring the student plan Google forcefully refunded, support is asking for:
• Full screenshot of the original SheerID verification email (with sender & recipient visible)
• Full screenshot of the cancellation & refund email
• Google Account Age Verification page screenshots
• Physical location, university name & university email
Has anyone actually gotten their student plan restored after sending all this??
7
2
58
30,259
Tibo's global Codex resets are starting to feel like a trap.
You carefully manage your weekly quota, leave some usage in the tank for important work, and then a “free reset” wipes everything you saved and restarts the entire 7-day timer.
The higher your plan, the more you lose.
People on $200/month plans are getting punished the hardest for being careful.
Just give us banked resets.
Let us keep what we didn’t use instead of forcing everyone to burn or lose it.
8
1
19
2,682
Microsoft accidentally leaked gpt-6.1-terra inside a GitHub Copilot SDK eval config.
Look closely at lines 6–10 in this config:
• Line 6: model: gpt-6.1-sol
• Line 7: judge_model: gpt-6.1-terra
• Line 10: executor: copilot-sdk (Microsoft’s internal staging router)
gpt-6.1-terra is already live inside Microsoft staging.
Microsoft engineers just leaked OpenAI's 'gpt-6.1-terra' by accident.
the official .NET EF Core repo has gpt-6.1-terra sitting in 18 evaluation configs.
17 of them were quietly upgraded from gpt-5.6-terra.
9
1,561
Google AI Pro student crackdown just got worse.
• People with prepaid balances through 2027 only got ~$5 refunded the rest of their credit vanished
• Some are re-subscribing to the $4.99 student plan without re-verifying
• Half of those accounts are already getting flagged
• Gemini 4 Argon is missing on the new student plans inside Antigravity
If your plan got wiped, check the refund amount and move fast before Google closes the door.
5
2
90
15,594
🚨 Google is internally testing an unlocked cybersecurity variant of Gemini 4 Argon (`gemini-4-argon-cyber-permissive`) alongside a new "Barium" model family.
Cyber & Routing:
• gemini-4-argon-cyber-permissive
• gemini-3.8-flash-router
• gemini-4-argon-disabled
New "Barium" Teamfood (-tf) Family:
• barium-b-concise-final-response-di-tf
• barium-b-final-visibles-v7-tf
Internal Frontier Eval:
• sonnet-5.5-high-bon-le (Sonnet 5.5 High Best-of-N)
Argon, Carbon, and now Barium are all inside Google's internal registry
7
4
70
13,161
Google is mass-cancelling student AI Pro upgrades this morning and force-refunding them.
Students who used the official upgrade button (or stacked credits) to move from the free 1-year AI Plus plan to AI Pro are waking up to:
• Unrequested refunds
• AI Pro plan wiped
• Original student plan also gone
• Back on the free 15 GB tier
Some are already hitting Gmail limits because their storage just vanished overnight.
Support is telling people the cancelled student promos can’t be restored.
Did your plan survive the purge?
22
12
293
35,435
🚨 Gemini 4 Carbon (`gemini-4-carbon`) has officially entered Google internal teamfooding.
New identifiers added in the drop:
• gemini-4-carbon
• gemini-4-carbon-teamfood
• carbon-g-medium-tf
• gemini-4-argon-teamfood
• gemini-4-argon-max-mode-tf
Carbon + Argon Max Mode are both live internally 👀
15
24
395
30,652
Holy shit.
Moonshot just admitted they had to nerf Kimi K3.1 Pro’s hacking ability before releasing the open weights.
An uncleaned checkpoint hit 90.5% on real kernel and Node.js exploits.
After scrubbing the data, it still scores 81%.
Open weights drop November 4.
2
12
828
Kimi K3.1 Pro vs the closed frontier:
Intelligence
• 55 on Artificial Analysis (4th overall)
• Beats GPT-6 Astra and Claude Fable 5.1
• Only 3 points behind Opus 5.5
Cost
• $2 / $10 per million tokens
• Roughly half the price of Opus 5.5
• Same price tier as Sonnet 5.5 and GPT-6.1 Sol
1
2
16
885
Gemini 4 Argon High usage check inside Google Antigravity
Look at the weekly timer: "Resets in 7m" with 67% Weekly Limit STILL remaining.
How much of your weekly limit do you have left today??
16
4
110
19,712
Kimi K3.1 Pro pricing:
Input: $2 / 1M tokens
Output: $10 / 1M tokens
Cached Input: $0.20 / 1M tokens
Cache Write: $2 / 1M tokens
4
13
1,030
Kimi K3.1 Pro launch night and the quota is already a disaster.
People are burning their entire 5-hour window in under 10 minutes.
Some in less than 2.
Thinking is forced on. Highspeed doubles the burn rate. One long prompt and you’re done.
Frontier benchmarks mean nothing when the model locks you out before it can finish a single real task.
This is the most aggressive usage limit I’ve seen from any major lab this year.
6
22
1,201
Kimi K3.1 Pro is out
You can go on the website and check yourself.
I am testing right now
5
31
2,257
$200 for Claude. $200 for ChatGPT.
A massive difference in AI usage value.
SemiAnalysis tested Claude and ChatGPT subscription limits and found Anthropic delivers roughly 5× the API-equivalent value on comparable mid-tier workloads.
That's a striking gap for two plans at the same price.
7
22
1,666
Gemini 4 Carbon is testing in arena
Built a 100% procedural 3D pelican riding a vintage beach cruiser in Three.js
2
1
32
1,656
I have never burned through Opus 5.5 this fast.
same tasks, same effort, and my session bar is gone in half the time it used to take.
is the token multiplier different or am I losing my mind?
anyone else watching Opus 5.5 torch their limit today?
19
5
131
4,184
Microsoft engineers just leaked OpenAI's 'gpt-6.1-terra' by accident.
the official .NET EF Core repo has gpt-6.1-terra sitting in 18 evaluation configs.
17 of them were quietly upgraded from gpt-5.6-terra.
13
6
90
11,870
Fable's own quota is gone.
It now eats your weekly limit.
Flawless.
Fable 5.5 is coming.
28
1
101
11,729
Claude Opus 5.5 started burning your quota in 30 minutes. now Codex is doing it in 15.
a single session on GPT-6 Astra at MEDIUM effort just drained the usage bar from 100% to 0% in ~15 minutes.
same model at MAX effort used to last 50+ minutes.
that's not a usage bar anymore, that's a countdown.
both labs quietly tightening token multipliers this week, or are the prompt caches just breaking?
11
3
52
4,217
Kimi K3.1 is about to release.
Cost: likely in the $3–5 / $15–25 range per million tokens
2
23
1,187
Opus 5.5 burned my weekly quota in 30 minutes.
This tweet is the only thing I got out of it.
19
8
224
6,948
Did Claude just rob me of an entire day of my weekly limit???
My reset has been Friday for months. today it says Saturday.
Nobody announced anything, my plan didn't change, it just silently moved up a day.
And i'm not alone. At 16:00 UTC the banked free reset expired and right then people started flooding in with the same story: reset day jumped a day early. Did you check yours?
11
4
88
3,248
Opus 5.5 usage check
Last week: felt unlimited.
Today: 100% of a Max (5x) session gone in 33 minutes.
Now locked out for 4 hr 27 min.
Anyone else?
102
23
659
32,511
Gemini 4 Argon
AI agent swarms just uncovered a 12-year-old 9.2-CVSS zero-day in core PHP (PHP2Shell / CVE-2026-109618) while offensive bot swarms race to exploit unpatched servers.
I built an interactive 3D simulation of the battle in a single HTML file:
1
10
882
