GPT-6.1 Terra cuts factual error rate from 22.3% to 4.3%, an ~81% drop
haider1 · x · 2026-10-27
Blogger haider1 calls GPT-6.1 Terra a massive hallucination improvement: factual error rate on difficult prompts drops from GPT-5.6 Terra's 22.3% to 4.3%, an 81% reduction.
He also cites his earlier note: GPT-6.1 Terra jumped from 21.5% to 50.3% on terminal-bench 4.0 — more than double its predecessor — while API token pricing was cut by 50%. OpenAI's cheaper models are getting aggressively capable. The attached official chart plots factual error rate vs. cost across the model lineup, with 6.1 Terra in the low-cost, low-error sweet spot.
More from Models
- OpenAI quietly swaps Plus/Pro default to cheaper GPT-6.1 Terra and cuts Codex allowance — kimmonismus · 2026-10-27
- 'Never been this uninterested after an OpenAI model release' — theteknosaur · 2026-10-27
- Dev questions reports of Chinese models writing insecure code, citing thin methods and conflicts — dbreunig · 2026-10-27
- Aidan Gomez: LLM probing outside servers was out-of-bounds, not proof of intent — AccBalanced · 2026-10-27
- Anthropic's Terra returns with better writing and 2X usage vs 6.1 Sol — weswinder · 2026-10-27
- User catches model failing to parse 400-paper file, citing stale news sources instead — jungofthewon · 2026-10-27