Updated April 2026

A year ago everyone used ChatGPT.
I don't use it anymore.

This isn't something that changes every week. It changes every day. New models, new prices, new tools — and what was standard 6 months ago can easily be outdated today.

My advice: stop trying to follow everything. Pick one tool that fits how you work and go deep with it. You'll switch again at some point — and that's completely normal.

Personal use

What I actually use

GitHub Copilot CLI is what I use the most right now. Not because it has the sexiest UI — it doesn't. But it sits right in the terminal, costs the same regardless of what Anthropic decides next quarter, and it just works. I can live with a slightly boring interface.

Claude (Anthropic) is what I use for thinking out loud. Architecture, security questions, explanations. But the free version is pretty limited at this point and prices keep going up. I pay for Copilot and use Claude for what can't be done in a terminal.

A year ago I only used ChatGPT. Now I barely touch it. Find your own way — it'll change anyway.

⚠️ Prices are going up — free tiers are shrinking

Most of these tools ran large free tiers in 2024-2025 while they chased users. That phase is over. Anthropic, OpenAI and Cursor have all cut what free users can do. Expect to pay if you want to use them seriously — and the prices in the table below could easily be wrong in three months.

Tool Best for Price from Free tier
Claude Thinking, analysis, long contexts $20/mo Yes (heavily limited)
GitHub Copilot IDE autocomplete, CLI agent, VS Code $10/mo Yes (limited)
Cursor Whole codebase as context $20/mo Yes (very limited)
Lovable Full app from prompt, fastest MVP $20/mo Yes (very limited)

Prices April 2026 — check official sites for current info

Current situation SUSPENDED 12 June 2026

The Fable 5 situation is not just news. It is an argument against vendor lock-in.

On 9 June 2026, Anthropic released Claude Fable 5 and Mythos 5. Three days later, both models were gone — switched off for all customers globally following an export control directive from the US government. No advance warning. No technical basis shared publicly. Just an order at 5:21pm ET and a deadline to comply.

This is the most concrete demonstration yet of what AI vendor lock-in actually means in a geopolitical context. You built workflows on a model. It is gone. Not because the company went bankrupt, not because they changed their pricing — but because a government made a decision and a commercial product disappeared for hundreds of millions of users with three days' notice.

Anthropic is a US company. That means it is subject to US export controls, US courts, and — as February 2026 demonstrated — threats of being declared a "supply chain risk" by the Pentagon if it refuses to grant the military unrestricted access. OpenAI is the same. Google is the same. All the major frontier models are American, and that is not without consequences for the rest of the world.

What this concretely means for you

  • Critical workflows built on a single cloud model are a single point of failure with geopolitical exposure
  • Open weights models (Llama, Mistral, Qwen) can be run locally — no one can switch them off
  • EU-based alternatives exist, but remain well behind the frontier in capability
  • Diversifying across providers is not just good practice — it is risk management

What is confirmed in the Fable 5 case

  • Confirmed: Fable 5 and Mythos 5 suspended 12 June 2026 for all users globally
  • Confirmed: Export control directive from the US Department of Commerce — not an Anthropic decision
  • Confirmed: Stated reason: a claimed jailbreak that can identify security vulnerabilities in code
  • Confirmed: Anthropic received no technical documentation and disputes that the finding justifies suspension
  • Confirmed: All other Claude models (Opus 4.8, Sonnet 4.6 etc.) continue to work normally
  • Context: Ongoing legal dispute between Anthropic and the Pentagon since February 2026 over military use — still before the courts

Source: Anthropic official statement, 12 June 2026 · Wikipedia: Claude (language model)

⌨️

Cursor

The closest you get to an AI that actually understands your project as a whole. Cursor indexes the entire codebase and can refactor across files in one go. Built on VS Code — all your extensions still work.

Codebase context Composer VS Code base Multi-file
FROM
$20/mo
Pro
Try Cursor →
✓ Pros
Whole codebase as context
Composer for multi-file changes
Built on VS Code
✗ Cons
Pro tier needed for serious use
Can feel overwhelming at first
🚀

Lovable

From idea to running app in minutes. Not just code — Lovable deploys too. Best for the MVP phase and prototypes. If you have an idea you want to show someone fast, this is the quickest tool. Supabase integration is solid.

React Supabase Deploy MVP Prototypes
FROM
$20/mo
Try Lovable →
✓ Pros
From prompt to deploy in minutes
Supabase integration
Great for React apps and prototypes
✗ Cons
Loses the plot on large projects
Best for MVP — not complex backends

Latest AI news

Live feed from Anthropic, GitHub and The Verge AI. Updates automatically.

Loading news...
THREAT INTELLIGENCE

Know your threat picture now.

See live security threats that could affect you and your home network.

Open SecIntel free →