Claude peaked at Opus 4.6 and it's downhill since, says 2-year user switching to GPT 6
ConcertDependent8452 · reddit · 2026-09-16
A Reddit user with 2+ years on Claude argues Opus 4.6 was the peak: the model obeyed instructions, cost less, and actually got work done. Since then, every "smarter" release has produced worse output (barely readable in other languages), higher prices, and a habit of adding unrequested caveats.
They complain that the past year's wave of AI-influencer "skills" exist to fix problems Anthropic itself introduced, burning even more tokens, only to vanish after every context compaction. After running GPT 6 via a fresh Codex install on their projects with zero skills, they found it works out of the box with fewer iterations and lower cost than Fable 5.1. The only reason they're still on Claude: OpenAI's x20 subscription remains blocked for new purchases.
More from Models
- OpenRouter quietly launches stealth model Union Alpha with DeepSWE results — airesearch12 · 2026-09-16
- Researcher warns LLM evals are broken after reading 'truly horrible' model traces — IanArawjo · 2026-09-16
- OpenAI's 10,000-word essay on a self-evolving 'alien mind' sparks ownership clash with startup JoyIn — SarahAnnabels · 2026-09-16
- Jev is a smart classifier, not a frontier LLM — a contrarian analysis — DavideCrapis · 2026-09-16
- Jev debuts at $0.042/M input tokens with free output, claiming 20-200x speedup — hey_abusiddik · 2026-09-16
- Custom Midtrain Plus Own RL Matches Astra Max at Half the Inference Price — hsu_byron · 2026-09-16