How did ChatGPT change by March 2026? From GPT-5.4 Thinking and the end of GPT-5.1 to Shopping, File Library, and the reported adult mode pause, this piece gently maps the recent shifts and where ChatGPT may be heading next.
www.reimagines.jp/chatgpt-upda...
Posts by REIMAGINES(リイマジンズ)
How someone rejects a confession often reveals more than the answer itself. This piece breaks down the gap between sincere and insincere responses, why the way you were rejected hurts so much, and what to watch for before your next relationship.
www.reimagines.jp/kokuhaku-kot...
Anthropic’s recent updates make more sense when you stop reading them as isolated features.
Cowork, Claude Code, computer use, security, and the Vercept acquisition all point in the same direction: Claude is being pushed from a conversational AI toward a working AI.
www.reimagines.jp/canthropic-d...
Claude has changed a lot since February 2026, but the updates are spread across models, work tools, safety features, and everyday usability.
I put together a beginner-friendly summary of Sonnet 4.6, Cowork, Claude Code Security, memory, and more
www.reimagines.jp/claude-recen...
GPT-5.4 and GPT-5.3 are closer in name than in purpose.
A simple way to think about it is this: GPT-5.4 is better for heavier, more complex work, while GPT-5.3 Instant feels better for everyday conversations and lighter tasks.
www.reimagines.jp/gpt-5-4-vs-g...
If you feel sleepy, heavy, or oddly irritated every spring, it’s not “just you.” Early spring often stacks three loads: autonomic switching fatigue, big temperature swings, and pollen-related inflammation. Start with a 10-minute reset today.
www.reimagines.jp/spring-malai...
GPT-4o is gone from the menu, and “just use 5.2” doesn’t really fix that feeling of loss.
I wrote a piece that:
• maps what actually happened in mid-Feb (4o retirement, 5.3-Codex)
• explains how Model Spec quietly shapes behavior
www.reimagines.jp/gpt4o-retire...
We are slowly entering a world where the one who listens to us the most is not a human, but an AI.
It feels safe. It feels kind.
And that is exactly why model shutdowns can hurt so much.
www.reimagines.jp/ai-emotion-h...
Many people say “only niche blogs can grow,” but your misc blog is not dead at all.
If you define a clear point of view, reconnect existing posts, and give each platform a simple role, you can grow the blog you already have.
www.reimagines.jp/zakkiblog-gr...
GPT 4o is scheduled to disappear from the ChatGPT UI on February 13.
That feels less like a simple model update and more like losing a familiar partner.
In this article I break down
– what actually changes on that day
– why keep4o is trending again
www.reimagines.jp/gpt4o-end-ke...
Getting outranked by a strong domain after you publish? It’s not personal. The keyword is being targeted.
Stop treating rankings as the goal. Protect the payoff system: conversion, retention, and compounding assets.I wrote a mental model for staying unbreakable in the AI era
note.com/reimaginesjp...
AI personalization is a build, not a miracle. ChatGPT is improving recall from past chats, while Gemini’s Personal Intelligence aims to unify your Google context. I compared the experiences and shared a leveling-up framework
www.reimagines.jp/chatgpt-memo...
Google’s Gemini just got a lot more personal.
I broke down “Personal Intelligence” in plain language:
• what it actually does
• which data it can use (Gmail, Search, YouTube, Photos)
• how pricing works
• and how to think about privacy without panic
www.reimagines.jp/gemini-perso...
Passkeys explained in plain terms
It is not about fancy crypto, it is about where your passkeys live and how they sync
iPhone plus Mac usually works best with Apple keychain
Android plus Windows often fits Google Password Manager
www.reimagines.jp/passkey-guid...
I wrote a practical guide to ChatGPT Health: what it can/can’t do, why it’s not for diagnosis, how to prep questions for your doctor, privacy basics, and rollout caveats (it may not be available in Japan yet).
www.reimagines.jp/chatgpt-heal...
If you feel busy every day
but see little real progress
you may not have a time problem
you may have a visibility problem
Try keeping a simple one week time log
Just 4 categories
work
household
future investment waste
No fifteen minute precision
just honest daily note
www.reimagines.jp/jikanbo1week/
If your to do list never ends
and you feel worse each night
the problem is not you
It is the system
Instead of tracking more tasks
start blocking time
Pick three important actions for today
turn them into calendar blocks
and let unfinished items slide to tomorrow
www.reimagines.jp/todolistsizon/
Meta’s reported acquisition of Manus isn’t only about “better AI.” It’s about AI that completes workflows: intake → proposal → revisions → delivery → iteration. I explain what changes for creators and small teams, and what to prepare now. #AIAgents #MetaAI #Manus
www.reimagines.jp/meta-manus-a...
If you only use Google Calendar
to record meetings and deadlines
you will always feel behind
Try this instead
Set four colors for
future investment
work
daily operations
rest
Block your future investment and rest first
then fit everything else around them
www.reimagines.jp/google-calen...
Most meetings run long because of poor preparation
not because people are bad at discussing
Create a simple agenda the day before
with purpose topics and decisions to be made
www.reimagines.jp/kaigi-mijika...
Most people do not lack time
They just spend it instead of investing it
Scrolling, random videos, and aimless browsing are time spending
Learning, exercising, building a side project, nurturing relationships, and resting are time investing
www.reimagines.jp/time-tsukaik...
Dating apps shouldn’t drain your self-worth.
I made a practical safety toolkit: a quick diagnosis, type-based rules, an exit decision tree, and 100 copy-paste templates (soft → firm).
If something feels off, you can stop early—without guilt. Note link in profile.
note.com/reimaginesjp...
Google in 2025 is quietly becoming “AI everywhere” — inside Search, Gmail, Photos, Chrome, Pixel, and Workspace.
I pulled 40 tips from Google’s official AI update and organized them into 5 scenes:
・Learning & research (Gemini 3, NotebookLM, Simplify)
www.reimagines.jp/2025googleai...
On Dec 18 2025 OpenAI made something clear
AI is no longer just about capability
It is about behavior
The updated Model Spec
the new Under 18 Principles
and age based experience design
all point to a shift
I explained what changed and why it matters
www.reimagines.jp/openaimodel-...
Gemini 3 Flash isn’t trying to be the smartest AI.
It’s trying to be the most used one.
Speed, low friction, and daily usability over raw power. Interesting move by Google.
www.reimagines.jp/gemini-3-fla...
Can ChatGPT Images generate before-and-after visuals?
After testing it a bit, I realized the process matters more than the answer.
Since beauty is subjective, I stopped short of conclusions.
If it makes you curious enough to try, that’s probably the point.
www.reimagines.jp/chatgpt-imag...
GPT Image 1.5 reveals its true character when generating Japanese beauty.
Soft features, averaged faces, restrained bodies.
Not a flaw, but a design choice for safety and commercial use.
Understanding this makes the model far more usable.
www.reimagines.jp/gpt-image-1-...
When will ChatGPT allow adult content?
December rumors are misleading.
Based on official statements and reports, 2026 Q1 looks far more realistic.
This article breaks down facts, rumors, and how ChatGPT compares to Gemini and Grok.
www.reimagines.jp/chatgpt-adul...