// BENCHMARK

MBPP+ benchmark

AI model leaderboard for the MBPP+ benchmark. Compare how large language models score on MBPP+, see the full ranking, and understand what this AI benchmark measures. LongCat-Flash currently leads with 81.27%. EvalPlus extension of MBPP with ~35x more tests per problem.

Leaderboard

# Model Organization Score Variant Source
#1 LongCat-Flash Meituan Inc 81.27% pass@1 official ↗
#2 Pangu Pro MoE Huawei 80.2% instruct-plus-pass@1 official ↗
#3 DeepSeek-V3 DeepSeek 78.8% plus-cited-minimax-text official ↗
#4 Mistral Small 3.2 Mistral AI 78.33% instruct-pass@5 official ↗
#5 Qwen2.5 Instruct (72B) Qwen 77% instruct-plus-cited-minimax-text official ↗
#6 GPT-4o (Nov 2024) OpenAI 76.2% plus-cited-minimax-text official ↗
#7 Gemini 2.0 Flash Google DeepMind 75.9% exp-plus-cited-minimax-text official ↗
#8 DeepSeek-V3 DeepSeek 75.7% pass@1 official ↗
#9 Gemini 1.5 Pro Google DeepMind 75.4% 002-plus-cited-minimax-text official ↗
#10 Claude 3.5 Sonnet Anthropic 75.1% 1022-plus-cited-minimax-text official ↗
#11 Mistral Small 3.1 Mistral AI 74.63% instruct-pass@5 official ↗
#12 dots.llm1 Rednote 74.5% pass@1 official ↗
#13 Llama 4 Scout Meta AI 73.5% plus-pass@1-cited-pangu-pro-moe official ↗
#14 Llama 3.1-405B Meta AI 73% instruct-plus-cited-minimax-text official ↗
#15 MiniMax-Text-01 MiniMax 71.7% plus-from-hf-readme official ↗
#16 Granite-4.0-H-Small IBM 71% pass@1 official ↗
#17 Granite-4.0-H-Tiny IBM 69% pass@1 official ↗
#18 Nemotron-H 56B NVIDIA 67.2% base-0-shot official ↗
#19 Nemotron-H 47B NVIDIA 65.6% base-0-shot official ↗
#20 Granite-4.0-H-Micro IBM 64% pass@1 official ↗
#21 Nemotron-H 8B NVIDIA 59.52% base-0-shot official ↗

Frequently asked questions about MBPP+

What is the MBPP+ benchmark?

EvalPlus extension of MBPP with ~35x more tests per problem.

How is the MBPP+ benchmark scored?

MBPP+ is scored using the pass@1 (%) metric, where a higher score is better. The maximum achievable score is 100.000. GenAIList aggregates reported scores from model providers and papers into a single ranked leaderboard.

Which AI model scores highest on MBPP+?

As of the latest reported scores on GenAIList, LongCat-Flash achieves the highest result on MBPP+ with a score of 81.27%.

Is a higher MBPP+ score better?

Yes. On MBPP+ a higher score indicates better performance, so models near the top of the leaderboard are the strongest.