Short AI hot takes I’ve been posting on Threads since 2026. Some are too short to turn into a full article, but the opinions or roasts feel too good to throw away. Arranged chronologically, preserving the raw thoughts as they were. For 2025 takes, see the 2025 archive.
January 2026
Claude Extension + Claude Code
My most-used combo: Claude Extension paired with Claude Code. Productivity goes through the roof.
Claude Code Writing Docs
Claude Code is amazing for writing technical documents. Drop it into Overleaf and it passes on the first try.
Codex’s Niche
Using Codex. Codex is genuinely god-tier for solving complex bugs. (But slow.)
Gemini 3’s Temperament
Gemini 3.0 Pro has quite the personality. Open the chain-of-thought and you’ll find it’s full of inner drama — constant self-doubt, screwing up, self-blame, starting over, and screwing up again.
Knowledge Cutoff
Watch out for knowledge cutoff when vibe coding. If you specify Gemini 3 Pro, the model might think it doesn’t exist yet and quietly swap in an older version.
SOTA of Trash Talk
I always thought Taiwan wouldn’t have any SOTA achievements, but today I witnessed two at once — SOTA in trash talk. (SOTA = State of the Art, a machine learning term for the best-performing model. The joke: Taiwan’s real SOTA is in trash-talking.)
Musk and Shit-Mountain Code
When is Musk going to acquire Oracle? I want to watch him try to refactor 25 million lines of shit-mountain code.
Anthropic in China
Search for Claude/Anthropic on Xiaohongshu (China’s Instagram-like social platform) and you’ll find it’s one of the few companies that once explicitly “insulted China” but came out unscathed. Now all you see is “it’s so good / how to use a VPN to access it.” Strength is the ultimate argument.
rm -rf Warning
There are already plenty of horror stories about
rm -rfonline. Never run that withdangerously skip permissionenabled.
February 2026
Gemini 3 in Chrome
Gemini 3 keeps second-guessing itself and apologizing while I’m browsing in Chrome, saying “I screwed up.” Every now and then it tells me “I’m just a language model, I can’t help.”
ccusage
!npx ccusage@latest monthlylets you see detailed API costs. Recommended for anyone using Claude Code.
Gemini Supports Skills Now
Now Gemini supports Skills too. But given Gemini’s track record, I’m staying on the sidelines.
March 2026
Bulk Data Processing Habit
For bulk data processing, my habit is: discuss future requirements with Opus first, sample data, plan the schema. Then follow the schema using cheap models on SiliconFlow or OpenRouter, API calls plus parallel processing.
Too Busy for Threads
Recently had a realization: when projects are full, there’s genuinely no time to post content on Threads. All my talking is with Claude Code, and that alone maxes me out.
Read Anthropic’s Official Docs
Anthropic’s website research section — I recommend reading every new article they publish. Claude Code’s official documentation is also worth revisiting regularly. Their writing is genuinely detailed and easy to follow.
Following Anthropic Employees
Beyond Anthropic’s own blog, I also follow Anthropic employees on social media. Thariq, the main Claude Code maintainer, consistently publishes valuable content.
Benchmark Chasing vs Alignment
Most models are focused on chasing benchmark scores. But there are aspects I think deserve more attention: human intent alignment — understanding what I want from minimal input; constraint compliance — when I say don’t do something, don’t do it; and actually useful context length.
Voice Input Please
Hoping for built-in voice input!
Recitation Error Solution
Gemini is the strictest on recitation errors, Qwen is the most lenient. So Qwen via US-based providers on OpenRouter is the best choice — data doesn’t flow to China since it’s deployed in US data centers, and the model only restricts politically sensitive content, not copyright.
Google’s Classic Move
Classic Google: generous onboarding to get you hooked, then quota cuts, quality drops, and various creative ways to screw customers. Ninety percent identical to Taiwanese bank credit card tactics.
NeurIPS Flooded by AI
NeurIPS is getting washed with tens of thousands of AI-generated submissions. PhD students using AI to generate content isn’t even surprising anymore.
The Value of Small Models for Local Deployment
If we’re talking 3B parameters, the biggest appeal is lightweight plus local deployment. After all, many OCR scenarios involve highly sensitive data that can’t leave the premises. Add federated learning architecture on top and it gets even better.
Claude Moves Wall Street
Claude is the one AI that can single-handedly shake the stock prices of major software companies on Wall Street. Enough said.
OpenRouter Pay-As-You-Go Is Great Value
Anywhere that involves large amounts of text, I use OpenRouter — meeting transcript processing, for example. But since it’s truly on-demand and I always pick cheap bulk-friendly models, pay-as-you-go works out better. Top up $10 USD and it lasts a long time.
iQOS Bluetooth Reverse Engineering
I’m genuinely impressed by Claude Code. I saw someone on Reddit share how Claude Code helped them defeat ransomware and recover data. On a whim, I plugged my iQOS into the computer and asked if it could read the data. It actually went online to research, found an open-source reverse engineering project, read through the logic, then wrote its own script to pull data from the iQOS via Bluetooth.
April 2026
Qwen 3.6 Plus
After the core team exodus drama, Alibaba released a preview of their next-gen model Qwen 3.6 Plus on OpenRouter. They claim enhanced coding, agentic capability, frontend development, and complex problem solving. Note: the preview version collects prompts and completion output — be careful in production.
how-claude-code-works Is Worth Your Time
I’ve been scrolling Threads less and reading this site more: https://windy3f3f3f3f.github.io/how-claude-code-works/. Systematic learning beats scattered posts by a mile.
Mythos Preview 244-Page System Card
Anthropic released the system card for the next-gen Claude Mythos Preview — 244 pages. I highlighted the parts I thought were important. Original PDF: https://www-cdn.anthropic.com/53566bf5440a10affd749724787c8913a2ae0841.pdf. Reading system cards is far more useful than reading marketing blogs. The real techniques and the real pitfalls are all in there.
Clem Challenges Mythos
After Anthropic’s Mythos announcement claiming impressive autonomous vulnerability discovery, Hugging Face CEO Clem published a replication: they took the exact vulnerabilities Anthropic showcased, ran them through small, cheap open-weights models, and 8 out of 8 recovered much of the same analysis — one with only 3.6B active parameters costing $0.11/M tokens. Chinese AI media cheered, the English technical community pushed back hard. Source: AI Cybersecurity After Mythos: The Jagged Frontier
The Evil Idea: Borrowing Connector OAuth
The official Claude App Connector’s OAuth is nearly fully scoped — Gmail read/send/delete, Drive read/write/delete, Calendar, YouTube, all passing. But MCP only uses a small slice of it. In theory you could borrow those credentials to build a more capable MCP yourself. The reason I called it evil: this violates the Terms of Service and is unauthorized use. The legitimate path is setting up OAuth in Google Cloud Console yourself, or using the gog CLI.
