Benchmark
FACTS Grounding Benchmark
What is FACTS Grounding?
A benchmark evaluating language models' ability to generate factually accurate and well-grounded responses based on long-form input context, comprising 1,719…
- Modality
- text
- Evaluates
- reasoning, factuality, grounding
- Openness
- unknown
- Importer
- LLM Stats
- Reported scores
- 13
Source provenance
FACTS Grounding results and reported scores
LLM Stats
| Model | Organization | Reported value | Reported | Evidence |
|---|---|---|---|---|
| Gemini 2.0 Flash | 0.836 | 2024-12-01 | evidence | |
| Gemini 2.0 Flash-Lite | 0.836 | 2025-02-05 | evidence | |
| Gemini 2.5 Flash | 0.853 | 2025-05-20 | evidence | |
| Gemini 2.5 Flash-Lite | 0.841 | 2025-06-17 | evidence | |
| Gemini 2.5 Pro Preview 06-05 | 0.878 | 2025-06-05 | evidence | |
| Gemini 3 Flash | 0.619 | 2025-12-17 | evidence | |
| Gemini 3 Pro | 0.705 | 2025-11-18 | evidence | |
| Gemini 3.1 Flash-Lite | 0.406 | 2026-03-03 | evidence | |
| Gemma 3 12B | 0.758 | 2025-03-12 | evidence | |
| Gemma 3 1B | 0.364 | 2025-03-12 | evidence | |
| Gemma 3 27B | 0.749 | 2025-03-12 | evidence | |
| Gemma 3 4B | 0.701 | 2025-03-12 | evidence | |
| GLM-5V-Turbo | Z.ai | 0.586 | 2026-04-02 | evidence |
Scores are partitioned by the source that reported them and are never merged into a single cross-source ranking, because the sources measure different things and say so.