reasoning benchmark
HumanEval Leaderboard
HumanEval is a benchmark for measuring functional correctness in programs synthesized from docstrings.
Updated Sep 5, 2026
Models66
Model coverage66
MetricScore
EvidenceB
HumanEval Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelOP | Score95.12% | Percentile100.00% | Participants66 | EvidenceC | Evaluated |
| Rank02 | ModelMA | Score94.50% | Percentile98.46% | Participants66 | EvidenceC | Evaluated |
| Rank03 | ModelAN | Score93.70% | Percentile96.92% | Participants66 | EvidenceC | Evaluated |
| Rank04 | ModelOP | Score93.40% | Percentile95.38% | Participants66 | EvidenceC | Evaluated |
| Rank05 | ModelMA | Score93.30% | Percentile93.85% | Participants66 | EvidenceC | Evaluated |
| Rank06 | ModelAC | Score92.70% | Percentile92.31% | Participants66 | EvidenceC | Evaluated |
| Rank07 | ModelOP | Score92.40% | Percentile90.77% | Participants66 | EvidenceC | Evaluated |
| Rank08 | ModelSA | Score92.10% | Percentile89.23% | Participants66 | EvidenceC | Evaluated |
| Rank09 | ModelAN | Score92.00% | Percentile87.69% | Participants66 | EvidenceC | Evaluated |
| Rank10 | ModelMA | Score92.00% | Percentile86.15% | Participants66 | EvidenceC | Evaluated |
| Rank11 | ModelAC | Score91.50% | Percentile84.62% | Participants66 | EvidenceC | Evaluated |
| Rank12 | ModelOP | Score90.20% | Percentile83.08% | Participants66 | EvidenceC | Evaluated |
| Rank13 | ModelIB | Score89.73% | Percentile81.54% | Participants66 | EvidenceC | Evaluated |
| Rank14 | ModelIB | Score89.73% | Percentile80.00% | Participants66 | EvidenceC | Evaluated |
| Rank15 | ModelGO | Score89.60% | Percentile78.46% | Participants66 | EvidenceC | Evaluated |
| Rank16 | ModelDE | Score89.00% | Percentile76.92% | Participants66 | EvidenceC | Evaluated |
| Rank17 | ModelME | Score89.00% | Percentile75.38% | Participants66 | EvidenceC | Evaluated |
| Rank18 | ModelAM Nova ProAmazon | Score89.00% | Percentile73.85% | Participants66 | EvidenceC | Evaluated |
| Rank19 | ModelME | Score88.41% | Percentile72.31% | Participants66 | EvidenceC | Evaluated |
| Rank20 | ModelMA | Score88.41% | Percentile70.77% | Participants66 | EvidenceC | Evaluated |
| Rank21 | ModelXA | Score88.40% | Percentile69.23% | Participants66 | EvidenceC | Evaluated |
| Rank22 | ModelME | Score88.40% | Percentile67.69% | Participants66 | EvidenceC | Evaluated |
| Rank23 | ModelAC | Score88.40% | Percentile66.15% | Participants66 | EvidenceC | Evaluated |
| Rank24 | ModelAC | Score88.40% | Percentile64.62% | Participants66 | EvidenceC | Evaluated |
| Rank25 | ModelAN | Score88.10% | Percentile63.08% | Participants66 | EvidenceC | Evaluated |
| Rank26 | ModelOP | Score88.10% | Percentile61.54% | Participants66 | EvidenceC | Evaluated |
| Rank27 | ModelOP | Score88.00% | Percentile60.00% | Participants66 | EvidenceC | Evaluated |
| Rank28 | ModelGO | Score87.80% | Percentile58.46% | Participants66 | EvidenceC | Evaluated |
| Rank29 | ModelOP | Score87.20% | Percentile56.92% | Participants66 | EvidenceC | Evaluated |
| Rank30 | ModelOP | Score87.10% | Percentile55.38% | Participants66 | EvidenceC | Evaluated |
HumanEval Highlights
The leading models and scores on this benchmark.
HumanEval Score Distribution
A closer view of the leading scores on this benchmark.
The Top AI Models for HumanEval
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
What is HumanEval?
What HumanEval measures and how its scores work.
HumanEval consists of 164 original programming problems.
It measures functional correctness and assesses language comprehension, algorithms, and simple mathematics.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- HumanEval
- Modality
- text
- Primary category
- reasoning
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- overall
LLM Stats. Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about HumanEval.
Which model scores highest on HumanEval?
MiniCPM-SALA is currently ranked first with 95.12%.
What are the top three models on HumanEval?
The current leaders are MiniCPM-SALA (95.12%), Kimi K2 0905 (94.50%), and Claude 3.5 Sonnet (93.70%).
Which HumanEval model has the lowest official input price?
Nova Micro has the lowest matched official input price at $0.04 input / $0.14 output per 1M tokens.
Which models are fastest among HumanEval results?
The fastest matched records are Llama 3.3 70B Instruct (2,220.00 tok/s via Cerebras), Llama 3.1 8B Instruct (2,047.00 tok/s via Cerebras), and Llama 3.1 70B Instruct (1,204.00 tok/s via Cerebras).
Does the highest score result prove overall model quality?
No. This benchmark measures one defined capability or task. The overall LLMBoard score uses a separate aggregation across eligible benchmark evidence.