CultureTalk-ID - Translation from Local Languages: leaderboard
CultureTalk-ID (2026) tests cultural commonsense in dialogues grounded in regional Indonesian cultures, in Indonesian and in local languages such as Javanese, Minangkabau and Balinese. This board scores zero-shot translation of the dialogues from the local languages into Indonesian with BLEU-4 (0 to 100); higher is better.
Metric: BLEU-4 (0-100) of zero-shot dialogue translation from the local languages into Indonesian with province and language context, mean of three runs for open-weight models; higher is better. Source: arxiv.org. Saturation forecast: Around March 2027. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Flash (Preview) | 52.74 | #144 |
| 2 | GPT-5.1 | 50.54 | #229 |
| 3 | Command A | 42 | #1274 |
| 4 | gemma2-9B-cpt-sahabatai-v1-instruct | 39.2 | #1488 |
| 5 | Gemma 2 9B (IT) | 30.41 | #2262 |
| 6 | Llama 3.1 8B Instruct | 26.6 | #4246 |
| 7 | Qwen 3 8B | 24.75 | #1830 |
| 8 | Sailor2-8B-Chat | 22.84 | #2396 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=culturetalk-id-translation-from-local-languages · How It Works · Data refreshed daily, snapshot 2026-11-13.