CanMT: leaderboard
Metric: Mean translation score (1-7) over the 12 CanMT directions (unweighted mean of the direction scores), GPT-5-nano judge with the reference translation scoring contextual accuracy, cultural adaptation, functional equivalence, fidelity and naturalness on 7-point Likert scales, averaged per sentence; default translation prompt; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 22 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Grok 4.1 | 5.18 |
| 2 | GPT-4 | 5.14 |
| 3 | DeepSeek V3.2 | 5.08 |
| 4 | DeepSeek R1 | 5.02 |
| 5 | Gemini 2.5 Flash Lite | 5 |
| 6 | GPT-4o | 4.96 |
| 7 | Qwen 2.5 72B Instruct | 4.91 |
| 8 | Llama 3.3 70B Instruct | 4.78 |
| 9 | Qwen 3 32B (Non-reasoning) | 4.74 |
| 10 | Qwen 3 14B (Non-reasoning) | 4.73 |
| 11 | Qwen 2.5 32B Instruct | 4.69 |
| 12 | Qwen 2.5 14B Instruct | 4.51 |
| 13 | Qwen 3 8B (Non-reasoning) | 4.48 |
| 14 | Qwen 2.5 7B Instruct | 4.1 |
| 15 | Qwen 3 4B (Non-reasoning) | 4.07 |
Interactive version: theaggregate.ai/benchmark?slug=canmt · How It Works · Data refreshed daily, snapshot 2026-10-07.