opus 5 is so brutally verbose
it always ends with some "while i was in there, i also noticed..." rant about a completely unrelated problem, explained in the most confusing way possible
even worse, it'll sometimes dump that unrelated garbage into the actual document
opus 5 consistently taking 10x the time to do simple tasks compared to 4.8. frustrating to come back an hour later & nothing has happened but 50% of token limit is gone
Opus-level intelligence running fully local at home on dual RTX 4090s.
Qwen3.8-27B-GGUF:
- 80 tok/s
- Full 262k context
- MTP on
- Only 34 GB of 48 GB VRAM used
- Outperforms Claude Opus max on SWE-Pro: 61.7 vs Claude Opus 4.6’s 53.4 GPQA: 89.2 vs 91.3
No need API bills, Or no rate limits.
Local frontier-class models are no longer theoretical.
Opus 5 has completely lost the plot recently. It leaves essays in comments in a way it never did before, recalling the entire provenance of a feature. Unhinged behavior.