Benchmarks
347 sources · 1,258 models × 3,038 benchmarks · 158,912 scoresAI Singapore's SEA-HELM (SouthEast Asian Holistic Evaluation of Language Models) tests models in seven Southeast Asian languages, Burmese, Filipino, Indonesian, Malay, Tamil, Thai and Vietnamese, with every prompt written in the language and posed zero-shot to instruct and reasoning models. Its pillars are NLP classics (understanding, generation and reasoning), LLM-specific tasks (SEA-IFEval instruction following and SEA-MT-Bench multi-turn chat), Southeast Asian linguistics (LINDSEA), culture and safety, plus a knowledge task. Each model runs 8 times; each task is scored 0-100 after rescaling against its random baseline and flooring at 0, a language's score is the unweighted mean of its competency scores, and the SEA-HELM score is the unweighted mean of the seven language scores. These are the scores of AI Singapore's leaderboard under its 8-run, seven-language protocol with the question-bootstrap aggregation SEA-HELM's repository published on August 5, 2026.
682nd strongest of 798 fitted benchmarks
Signal-to-noise +2.69
609th hardest of 798 fitted benchmarks
Capability needed 123.4
| # | Model | Setting | Score | Source |
|---|---|---|---|---|
| 1 | Gemini 3.1 Pro Preview | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 77.4 | |
| 2 | Gemini 3.8 Flash | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 77.3 | |
| 3 | GPT-5.6 Sol | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 77.0 | |
| 4 | GPT-6 Astra | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 76.9 | |
| 5 | Gemini 3.6 Flash | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 76.4 | |
| 6 | GPT-5.5 | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 76.3 | |
| 7 | Gemini 3.5 Flash | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 76.3 | |
| 8 | Gemma 4 31B IT | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 76.2 | |
| 9 | Qwen3.5 397B A17B | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 76.2 | |
| 10 | GPT-5.4 | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 76.0 | |
| 11 | Opus 4.8 | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 75.6 | |
| 12 | GPT-5.6 Terra | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 75.4 | |
| 13 | Gemini 3.1 Flash-Lite | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 75.3 | |
| 14 | Gemma 4 26B A4B IT | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 74.8 | |
| 15 | Qwen3.5 122B A10B | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 74.7 | |
| 16 | Qwen3.6 27B | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 74.6 | |
| 17 | Qwen3.5-27B | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 74.2 | |
| 18 | Sonnet 4.6 | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 74.1 | |
| 19 | Qwen3.6 35B A3B | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 73.9 | |
| 20 | Sonnet 5 | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 73.8 | |
| 21 | Qwen3.5 35B A3B | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 73.5 | |
| 22 | Qwen3.8-Flash-Next | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 73.0 | |
| 23 | GPT-5.4 Mini | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 72.8 | |
| 24 | MiniMax M3 | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 72.3 | |
| 25 | Muse Glimmer 30B | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 72.1 | |
| 26 | GPT-5.6 Luna | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 72.1 | |
| 27 | Inkling-Small | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 71.5 | |
| 28 | GLM-4.7 | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 71.5 | |
| 29 | Qwen3-VL-235B-A22B-Instruct | not supported · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); the model has no reasoning mode | 71.0 | |
| 30 | Gemma 4 12B IT | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 69.3 | |
| 31 | Qwen3.5-9B | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 69.1 | |
| 32 | Qwen3 32B | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 67.7 | |
| 33 | Qwen3 VL 32B Instruct | not supported · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); the model has no reasoning mode | 67.0 | |
| 34 | Gemma 4 E4B IT | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 66.3 | |
| 35 | gpt-oss-120b | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 66.1 | |
| 36 | Haiku 4.5 | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 64.9 | |
| 37 | Qwen3 14B | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 64.7 | |
| 38 | Mistral Medium 3.5 | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 64.7 | |
| 39 | Gemma 3 27B IT | not supported · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); the model has no reasoning mode | 64.5 | |
| 40 | Nemotron 3 Ultra 550B A55B | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 64.1 | |
| 41 | Qwen3.5-4B | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 63.8 | |
| 42 | Llama 4 Scout Instruct | not supported · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); the model has no reasoning mode | 63.4 | |
| 43 | Gemma 3 12B IT | not supported · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); the model has no reasoning mode | 62.4 | |
| 44 | Llama 4 Maverick Instruct | not supported · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); the model has no reasoning mode | 61.3 | |
| 45 | gpt-oss-20b | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 60.6 | |
| 46 | Llama 3.3 70B Instruct | not supported · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); the model has no reasoning mode | 60.4 | |
| 47 | Qwen3 8B | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 60.3 | |
| 48 | Gemma 4 E2B IT | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 58.8 | |
| 49 | Command A+ (May 2026, BF16) | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 56.7 | |
| 50 | Qwen3-VL-8B-Instruct | not supported · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); the model has no reasoning mode | 56.6 | |
| 51 | GLM 4.7 Flash | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 53.8 | |
| 52 | Mistral Small 4 | disabled · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning off (the board labels the row non-reasoning) | 52.1 | |
| 53 | Qwen3 VL 4B Instruct | not supported · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); the model has no reasoning mode | 50.4 | |
| 54 | Nemotron 3 Super 120B A12B | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 49.3 | |
| 55 | Gemma 3 4B IT | not supported · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); the model has no reasoning mode | 47.8 | |
| 56 | Nemotron 3 Nano 30B A3B | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 46.9 | |
| 57 | Nemotron 3.5 Lightning 30B A3B BF16 | thinking · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); reasoning on (the board labels the row reasoning); effort or thinking budget unpublished | 46.1 | |
| 58 | Olmo 3.1 32B Instruct | not supported · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); the model has no reasoning mode | 40.9 | |
| 59 | Llama 3.1 8B Instruct | not supported · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); the model has no reasoning mode | 37.2 | |
| 60 | Apertus 70B Instruct | not supported · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); the model has no reasoning mode | 32.7 | |
| 61 | Apertus 8B Instruct | not supported · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); the model has no reasoning mode | 27.5 | |
| 62 | Llama 3.2 3B Instruct | not supported · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); the model has no reasoning mode | 25.1 | |
| 63 | Tiny Aya Global | not supported · SEA-HELM leaderboard: zero-shot native-language prompts in seven Southeast Asian languages, 8 runs per model with the model's default generation settings (vLLM defaults otherwise); the model has no reasoning mode | 19.2 |
Benchmarks outside its family whose model rankings track it most reliably (Spearman ρ over at least 10 shared models).