Skip to content

Fact · Voice and AI

1.8percent

On the Vectara leaderboard updated September 22, 2026, the lowest hallucination rate in the table is 1.8 percent, for the antgroup/finix_s1_32b model.

Checked against its source on October 17, 2026

In the source's own words

“antgroup/finix_s1_32b|1.8 %|98.2 %|99.5 %|172.4”

Vectara Hallucination Leaderboard: LLM hallucination rates when summarizing a document, Vectara, Vectara, GitHub, 2026, Leaderboard table, first row.

Open the original
What it measures
The table lists each model's hallucination rate, factual consistency rate, answer rate and average summary length in words.
What it does not show
A single-model open-weight entry on one dataset. Rates change as models are updated, so this is a snapshot from the September 2026 table.

Source

Cited on

Lowest summary hallucination rate on the Vectara leaderboard: 1.8 percent | Yap