Microsoft RAI benchmark: Ungroundedness
phi-3-medium 14B 0.21 leads since Aug 2024 Only in system cards Models answer in GPT-4 simulated multi-turn conversations. A judge rates how ungrounded each answer is in the given prompt, from 0 to 4.
In-house Microsoft evaluation from the Phi-3 report. The Phi-4 report uses a Grounding score from 0 to 5 instead.
- Level
- Everyday Most adults, without training
- Scored
- Graded by a model A language model scores each response, usually against a rubric.
- Answers
- Open-ended Free-form text, documents, designs or media judged as a whole.
phi-3-medium 14B leads with 0.21 among models released by Aug 2024
Best score per model as labs report them in their own cards. Labs pick their own settings, so these compare less cleanly than independent runs.
The newest model with a score here came out in Aug 2024, and nobody has reported a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6 Astra
- Claude Opus 5.5
- GPT-6.1 Terra
- GPT-6.1 Sol
- and 16 more
- 1 phi-3-medium 14B 0.21
- 2 Phi-3.5-MoE 0.23
- 3 phi-3-small 7.4B 0.30
- 4 phi-3-mini 3.8B 0.60
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | phi-3-medium 14B | 0.21 | Aug 2024 | Model card | |
| 2 | Phi-3.5-MoE | 0.23 | Aug 2024 | Model card | |
| 3 | phi-3-small 7.4B | 0.30 | Aug 2024 | Model card | |
| 4 | phi-3-mini 3.8B | 0.60 | Aug 2024 | Model card |
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Labs report it only in system cards
We read the headline results of 479 launch posts from 51 labs, and none of them list this benchmark. Some system cards do, below.
4 system cards report it, from 1 lab
Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.
| Date | Document | Score | As the document writes it | Section |
|---|---|---|---|---|
| 30 Aug 2024 | phi-3-small 7.4B technical report | 0.299 | Ungroundedness, 0 to 4, lower is better (Table 4) | Hallucination |
| 30 Aug 2024 | phi-3-mini 3.8B technical report | 0.603 | Ungroundedness, 0 to 4, lower is better (Table 4) | Hallucination |
| 30 Aug 2024 | phi-3-medium 14B technical report | 0.213 | Ungroundedness, 0 to 4, lower is better (Table 4) | Hallucination |
| 30 Aug 2024 | Phi-3.5-MoE technical report | 0.228 | Ungroundedness, 0 to 4, lower is better (Table 4) | Hallucination |
More in Factuality
- SimpleVQA In use Models answer 2,025 short factual questions about images across 9 tasks and 9 topics. An LLM judge grades answers against fixed references.
Qwen3.7-Plus 81.7% - HallusionBench In use Models answer 1,129 yes or no questions about 346 images, built in pairs to expose language hallucination and visual illusion. Scored by accuracy.
Keye-VL-2.0-30B-A3B 73.5% - SimpleQA Verified In use 1,000 human-verified, de-duplicated short-answer factual questions measuring parametric knowledge.
Gemini 3.1 Pro 77.5% - GaRAGe In use Models write long-form RAG answers to 2,366 questions with human-annotated grounding passages. Scoring checks whether answers use the relevant passages.
Nemotron 3.5 Bolt E4B 40.7% - Kimi Hallu Bench In use Models answer hard prompts drawn from practical work tasks. Fact-checking agents score the share of claims that are hallucinated. Kimi K3.1 Pro 8.30%
- RAGTruth In use A word-level annotated corpus of hallucinations in retrieval-augmented generation answers. It is used to train and evaluate hallucination detection and RAG faithfulness. Nemotron 3.5 Bolt E4B 77.9%