Benchmarks
Adversarial NLI
Models scored
15
Best score
0.58
Metric
Score
Adversarial NLI is an AI evaluation tracked by Epoch AI, with 15 scored model versions on record. The scores here are reported by third parties rather than produced by Epoch AI running the evaluation itself.
Scored models span January 2022 to April 2024. The best score recorded is 0.58, measured as Score. Models come from Microsoft, Mistral AI, Google DeepMind, Meta AI, OpenAI, Microsoft,NVIDIA.
Full record
- Score metric
- Score
- Models scored
- 15
- Best score recorded
- 0.58
- Earliest model scored
- Jan 27, 2022
- Most recent model scored
- Apr 23, 2024
- Third-party reported
- Yes
Leaderboard
| 01 | gpt-3.5-turbo-1106 | OpenAI | 0.58 |
| 02 | Phi-3-small-8k-instruct | Microsoft | 0.58 |
| 03 | Meta-Llama-3-8B-Instruct | Meta AI | 0.57 |
| 04 | Phi-3-medium-128k-instruct | Microsoft | 0.56 |
| 05 | Mixtral-8x7B-v0.1 | Mistral AI | 0.55 |
| 06 | Phi-3-mini-4k-instruct | Microsoft | 0.53 |
| 07 | gemma-7b | Google DeepMind | 0.49 |
| 08 | Mistral-7B-v0.1 | Mistral AI | 0.47 |
| 09 | phi-2 | Microsoft | 0.42 |
| 10 | Megatron-Turing NLG 530B | Microsoft,NVIDIA | 0.4 |
| 11 | text-davinci-001 | OpenAI | 0.35 |
More benchmarks
SourceEpoch AI, 'AI Benchmarking Hub'. Published online at epoch.ai. Retrieved 2026-07-29 from https://epoch.ai/benchmarks. Licensed under CC BY 4.0.
← All benchmarks