MT-Bench
Multi-turn chat quality scored 1 to 10 by an LLM judge
knowledge and reasoning
Score
Definition
- Version
- not in the notes
- Category
- knowledge and reasoning
- Maintainer
- not in the notes
- Status
- not in the notes
- Home page
- not in the notes
Sources
- huggingface.co/nvidia/Nemotron-4-340B-Instruct https://huggingface.co/nvidia/Nemotron-4-340B-Instruct
- qwenlm.github.io/blog/qwen1.5-110b https://qwenlm.github.io/blog/qwen1.5-110b/
- mistral.ai/news/la-plateforme https://mistral.ai/news/la-plateforme/
Scores at a Glance
Score
Higher is better for this unit.
1
Qwen1.5-110B-ChatAlibaba (Qwen)average score
Mistral Medium (Dec '23)Mistral AI
Nemotron-4 340BNVIDIAInstruct model; judged by GPT-4-0125-Preview
Qwen1.5-110B-ChatAlibaba (Qwen)average score8.88 score
vendor-reported2Mistral Medium (Dec '23)Mistral AI8.6 score
vendor-reported3Nemotron-4 340BNVIDIAInstruct model; judged by GPT-4-0125-Preview8.22 score
vendor-reported02.557.510
Where this comes from
Each bar is one model, with its most independent recorded score. Hatched bars are vendor-reported; the word beside each bar says who ran it. Settings and dates differ between rows; the Scores table below has every result.
Scores
| Model | Score | Who ran it | Setting | View or population | Date | Source |
|---|---|---|---|---|---|---|
| Qwen1.5-110B-Chat | 8.88 score | vendor-reported | average score | – | 25 Apr 2024 | qwenlm.github.io |
| Mistral Medium (Dec '23) | 8.6 score | vendor-reported | – | – | 11 Dec 2023 | mistral.ai |
| Nemotron-4 340B | 8.22 score | vendor-reported | Instruct model; judged by GPT-4-0125-Preview | – | Jun 2024 | huggingface.co |
Where this comes from
Scores from different settings, harnesses or dates are not directly comparable; the row says which. Sorted by score, highest first, as a reading aid, not a ranking.
To cite this page: Project AURION, AI Use-Cases Research & Innovation, "MT-Bench", https://aurion.semai.id/benchmarks/mt-bench/, records of October 2026.