MBPP+
A cleaned 378-task MBPP evaluation with roughly 35 times more tests for stricter Python functional correctness.
The 2026-07-13 BenchmarkList snapshot tops out at 80.2% pass@1, while the small public task set, provider-sourced reporting, and dataset-version sensitivity limit clean frontier discrimination.
Performance Timeline
Longitudinal progression of model scores against human baselines.Performance & Historical Trajectory
Empirical score progression across model release dates and evaluation rounds.
| Model | Score | Date | Source Type | Provenance |
|---|---|---|---|---|
| Qwen2.5-Coder-32B-Instruct | 77% | 2024-11-12 | vendor-reported | Source ↗ |
| Gemini 1.5 Pro 002 | 74.6% | 2024-09-24 | vendor-reported | Source ↗ |
| O1 Preview (Sept 2024) | 80.2% | 2024-09-12 | vendor-reported | Source ↗ |
| O1 Mini (Sept 2024) | 78.8% | 2024-09-12 | vendor-reported | Source ↗ |
| DeepSeek-Coder-V2-Instruct | 75.1% | 2024-06-17 | vendor-reported | Source ↗ |
Human Baseline & Difficulty Horizon
Calibrated human reference points, specialist benchmarks, and ceiling thresholds.Human software engineer pass@1 solve rate under expanded edge-case contract test suites.
Metric & Scoring Methodology
Verification protocols, aggregation formulas, and specialized metric variants.MBPP+ pass@1 (%)Dataset & Compute Cost
Evaluation volume, public availability, API pricing, and local hardware requirements.$5 – $20 USD for full benchmark evaluation run on frontier APIs.
1x NVIDIA RTX 4090 (24GB) or A100 (40GB/80GB) via vLLM / SGLang
How to Run & Reproduce
Standardized evaluation protocols, CLI commands, and reproducible runner templates.lm_eval --model hf --model_args pretrained=<model_path> --tasks mbpp-plus --batch_size autoopencompass --datasets mbpp-plus --models <model_config># Standard API Evaluation Loop
from openai import OpenAI
client = OpenAI()
response = client.chat.completions.create(
model="gpt-4o",
messages=[{"role": "user", "content": prompt}],
temperature=0.0,
)When publishing results for MBPP+, always report the exact prompt template, few-shot exemplar ordering, sampling temperature (temperature=0), maximum reasoning budget tokens, and the precise timestamped model snapshot ID.
Contamination & Memorization Analysis
Audit of pretraining exposure risks, memorization vectors, and refresh policies.Static fixed snapshot
Public on web / HuggingFace