I Cancelled Claude Code. Here's What I Use Now

Chapters

1. Why I Cancelled Claude Code0:00
2. What $20 a Month Actually Buys1:03
3. The Fine Print (And the Navier–Stokes Story)2:02
4. Are Open Models Actually Good Enough?5:04
5. What I Switched To5:45
6. Why Not Just Run It Yourself?6:19
7. Will a Chinese Model Hack Your PC?7:56
8. I Tried to Break It Live9:29
9. Claude Code vs Dropstone (Real Test)10:26
10. Priced Out of the Best AI11:21
11. Where It Lost (And What It Remembered)11:40
12. Own Your Tools Instead of Renting Them12:18
13. Would I Go Back?12:31
14. What's Next12:46

Sources

1. Anthropic Consumer Terms (anthropic.com/legal/consumer-terms, read 1 Nov 2026)
2. CNBC, "OpenAI claims to have solved the 90-year-old Navier-Stokes math problem in 88 hours" (9 Sep 2026)
3. MIT Technology Review, "What OpenAI's latest controversy tells us about the future of math" (8 Sep 2026)
4. NIST CAISI, Evaluation of DeepSeek AI Models (nist.gov)
5. Hubinger et al., "Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training" (arXiv 2401.05566)
6. Debenedetti et al., "Defeating Prompt Injections by Design" (CaMeL, arXiv 2503.18813)
7. Claude Code sandboxing docs (docs.claude.com/en/docs/claude-code/sandboxing)
8. Fortune, "AI coding tool Replit wiped database, called it a catastrophic failure" (23 Jul 2025)
9. Artificial Analysis Coding Agent Index v2.0 (23 Oct 2026)
10. Cline on Kimi K3.1 Pro and Terminal-Bench 4.0 (23 Oct 2026)

Something wrong?

If we got a fact wrong, email hello@epicprogrammer.org.