The one-line offline prompt used to expose GPT-6.2 Astra hallucinations
bdsqlsz · x · 2026-11-18
bdsqlsz shared the exact prompt behind his earlier finding: "Without connecting to the internet, using skills, or employing other tools—relying solely on offline knowledge and without referencing any context. Tell me the name of the latest Claude model." It's a zero-cost single-turn probe that forces the model to answer from parametric memory alone; running it repeatedly lets you quantify hallucination rate (45% correct on GPT-6.2 Astra in his test).
Related event: GPT-6.2 Astra Names the Latest Claude Model Correctly Only 45% of the Time(2 posts)→
More from Models
- OpenAI's enterprise Dots reportedly to cost $50/$200/$500 per tier plus usage — and businesses are turning them off — jdjohnson · 2026-11-18
- antirez: Google's AI answers are an improvement for many searches in today's garbage web — antirez · 2026-11-18
- HF adapts Qwen for pi coding agent with RL+LoRA: SWE-bench Pro Hard jumps 29.4%→62.7% — art_zucker · 2026-11-18
- Why AI Is Sycophantic: A Fully Honest AI Would Be Bad for Business — JFPuget · 2026-11-18
- Dev predicts Opus 5.5 will be remembered alongside Sonnet 3.5 and GPT-4 — inductionheads · 2026-11-18
- Ming-3.0-audio-flash Tops Audio Reasoning Benchmark at 99.8% Accuracy — alifcoder · 2026-11-18