BenchLeader

FACTS Grounding leaderboard

As of 16 Nov 2026, the top score on FACTS Grounding is 84.2% by GPT-6.1 Terra (OpenAI), ahead of GPT-6 Sol at 83.3%, across 26 model configurations with a published result (Google, result dated 11 Nov 2026). Long-form answers written from a supplied document and nothing else: every claim has to be supported by the text, and an answer that does not address the request scores nothing. Published by Google. FACTS Grounding is shown for reference only and does not count toward the BenchLeader Index.

Key facts (as of 16 Nov 2026)

  • GPT-6.1 Terra holds the top FACTS Grounding score at 84.2% (Google, 11 Nov 2026).
  • GPT-6 Sol is second on FACTS Grounding at 83.3% (Google, 11 Nov 2026).
  • FACTS Grounding: Long-form answers written from a supplied document and nothing else: every claim has to be supported by the text, and an answer that does not address the request scores nothing. Published by Google (Google).
  • 26 model configurations have a published FACTS Grounding result on BenchLeader (BenchLeader, 16 Nov 2026).
  • FACTS Grounding does not feed the BenchLeader Index; it is shown for reference (BenchLeader methodology).

Long-form answers written from a supplied document and nothing else: every claim has to be supported by the text, and an answer that does not address the request scores nothing. Published by Google.

Published by
Googledata via Kaggle Benchmarks
Category
Long context
Index weight
Reference only
Models
26
Data as of
16 Nov 2026
Newest result
12 Nov 2026

Kaggle (kaggle.com); each board credits its publishing lab.

26 of 26
#
1GPT-6.1 Terra84.2%–2026-11-11
2GPT-6 Sol83.3%–2026-11-11
3GPT-6 Astra80.1%66.42026-10-24
4GPT-6.1 Sol78.0%62.02026-10-24
5Gemma 4 26B A4B77.8%56.12026-10-24
6Gemini 4 Carbon77.2%–2026-10-29
7GPT-6 Luna76.7%–2026-10-27
8Gemma 4 31B75.8%51.92026-10-24
9Qwen3.8 Maxmax73.8%58.72026-10-27
10Gemini 4 Argon73.3%64.42026-10-24
11Grok 4.768.6%–2026-10-24
12Gemini 3.8 Flash68.2%56.42026-10-23
13Gemini 3.6 Flash66.6%49.32026-11-11
14Grok 4.666.5%–2026-11-12
15Gemini 3.5 Flash Lite66.0%51.42026-11-11
16Claude Sonnet 5.562.5%–2026-11-12
17Claude Opus 5.562.5%61.02026-10-24
18GPT-5.6 Sol61.9%54.32026-11-11
19GPT-5.6 Terra60.5%–2026-10-24
20Claude Haiku 5.557.0%–2026-10-24
21Gemini 3.1 Pro56.6%59.92026-10-24
22GLM 5.3 Flash56.1%57.82026-10-27
23GLM 5.355.2%–2026-11-12
24GPT-5.6 Luna52.5%47.22026-11-11
25Gemini 3 Flash48.6%56.52026-11-12
26gpt-oss-120b20.1%46.12026-10-24

More long context benchmarks · All LLM benchmarks

Cite as: BenchLeader, “FACTS Grounding leaderboard”, data as of 16 Nov 2026, https://www.benchleader.com/benchmarks/kaggle_facts_grounding.

FACTS Grounding: questions

What does FACTS Grounding measure?
Long-form answers written from a supplied document and nothing else: every claim has to be supported by the text, and an answer that does not address the request scores nothing. Published by Google. Scores are reported in percent of tasks solved; higher is better.
What is the top score on FACTS Grounding?
As of 16 Nov 2026, the highest score on FACTS Grounding is 84.2% by GPT-6.1 Terra (result dated 11 Nov 2026), ahead of GPT-6 Sol at 83.3%. Scores come from Google and are re-read daily.
Which AI model leads FACTS Grounding?
GPT-6.1 Terra leads FACTS Grounding with 84.2% as of 16 Nov 2026 (result dated 11 Nov 2026), ahead of GPT-6 Sol at 83.3%.
How many models have FACTS Grounding results?
26 model configurations have a FACTS Grounding result on BenchLeader, all taken from Google via Kaggle Benchmarks.
Who runs FACTS Grounding and how often is it updated?
FACTS Grounding is published by Google. BenchLeader re-reads the published results once a day and records the date each result was published.
Does FACTS Grounding count toward the BenchLeader Index?
No. FACTS Grounding is shown for reference but left out of the composite index.