VQV Signal
According to a post on X by @HalachaBench, the project gave AI models from the makers of ChatGPT, Claude, Gemini and others real ordination exam questions from Israel's Chief Rabbinate. The questions were in Hebrew, and the models had no books or internet access. The best model scored 80 out of 100, while many scored under 11. The post also says open-weight models did worse, as noted by @dinabass.
According to a post on X by @HalachaBench, the project gave AI models from the makers of ChatGPT, Claude, Gemini and others real ordination exam questions from Israel's Chief Rabbinate. The questions were in Hebrew, and the models had no books or internet access. The best model scored 80 out of 100, while many scored under 11. The post also says open-weight models did worse, as noted by @dinabass.
AI-assisted summary based on the listed source.
HalachaBench gave models from the makers of ChatGPT, Claude, Gemini and others real ordination exam questions from Israel's Chief Rabbinate, in Hebrew, with no books or internet. The best scored 80 out of 100 and many scored under 11; @dinabass notes open-weight models did worse. Why it matters: It's a fresh test...
The results offer a fresh look at how well current AI models handle specialized reasoning in a language other than English, according to the post. Developers and businesses that rely on AI for non-English or niche professional tasks may see that performance can vary widely. Readers should check the full benchmark details before drawing firm conclusions.
VQV organizes public signals from inspectable sources. It does not independently verify the underlying report.
Signal Strength reflects source quality, relevance, freshness and evidence. Public Interest helps organize discovery; it is not proof of truth.
VQV surfaced this signal because it is recent, relevant to Consumer AI, connected to Scobleizer News Pulse (X).
No login, cookies, social SDKs, or automatic posting.
