SEA-HELM
(Southeast Asian Holistic Evaluation of Language Models)
Data last updated Nov 18, 2026
SEA-HELM is an assessment of large language models across various tasks, with an emphasis on Southeast Asian languages. The leaderboard evaluates models across key multilingual capabilities such as proficiency in Southeast Asian chat, instruction-following in Southeast Asian languages, Southeast Asian linguistic tasks and performance on a suite of English tasks.
80 open-weights models
Evaluated through local model deployment
21 closed-weights models
Evaluated through their respective APIs
Update
· Nov 17, 2026
- Added new evaluations: LINDSEA 2.0, KOPI-SEA
- Added new model results: SEA-LION v5 models, GLM 5.3, Nemotron 3.5, GPT 6, Gemini 4, and more.
- Added the Tokenizers page highlighting the number of vocabulary tokens each tokenizer spends on Southeast Asian scripts.
SEA Overall
Average of 2000 bootstraps. 95% CI are shown.
Performance for each SEA Language
Average of 2000 bootstraps. 95% CI are shown.
Legend:
Reasoning model
Model Size: All
Open models only