FACTS Grounding
A Google DeepMind benchmark that measures how factually grounded an LLM's long-form responses are to a provided source document, scoring the share of responses that are eligible and fully supported by the context with no hallucinations.
What this benchmark measures
A Google DeepMind benchmark that measures how factually grounded an LLM's long-form responses are to a provided source document, scoring the share of responses that are eligible and fully supported by the context with no hallucinations.
Rows on this page are sourced from public benchmark artifacts, leaderboard exports, or source-linked model reports. Each row keeps benchmark version, source model name, and available run details attached to the score.
The metric shown here is Grounding accuracy. It should be interpreted within FACTS Grounding, not compared as part of a site-wide ranking.
Frequently asked
What is FACTS Grounding?
A Google DeepMind benchmark that measures how factually grounded an LLM's long-form responses are to a provided source document, scoring the share of responses that are eligible and fully supported by the context with no hallucinations. It is a reasoning benchmark measured by Grounding accuracy.
What does Grounding accuracy mean on FACTS Grounding?
FACTS Grounding reports Grounding accuracy (%); higher is better. Scores are shown only within FACTS Grounding and are never averaged with other benchmarks.
What is the top reported FACTS Grounding score?
Gemini 2.0 Flash has the top reported score on FACTS Grounding: 83.6% (Grounding accuracy).
Why do FACTS Grounding scores differ across runs?
Harness, scaffold, reasoning effort, and prompt setup change results, so two runs of the same model can differ. evals.report keeps each score with its run context so the differences stay visible.
Does evals.report rank models across benchmarks?
No. FACTS Grounding scores are shown within their own metric; evals.report never combines benchmarks into a composite ranking or a single "best model".