reasoning benchmark
HumanEval Leaderboard
A benchmark that measures functional correctness for synthesizing programs from docstrings, consisting of 164 original programming problems assessing language comprehension, algorithms, and simple mathematics
Updated Aug 17, 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 | Score95.1% | Percentile100.0% | Participants66 | EvidenceC | Evaluated |
| Rank02 | ModelMA | Score94.5% | Percentile98.5% | Participants66 | EvidenceC | Evaluated |
| Rank03 | ModelAN | Score93.7% | Percentile96.9% | Participants66 | EvidenceC | Evaluated |
| Rank04 | ModelOP | Score93.4% | Percentile95.4% | Participants66 | EvidenceC | Evaluated |
| Rank05 | ModelMA | Score93.3% | Percentile93.8% | Participants66 | EvidenceC | Evaluated |
| Rank06 | ModelAC | Score92.7% | Percentile92.3% | Participants66 | EvidenceC | Evaluated |
| Rank07 | ModelOP | Score92.4% | Percentile90.8% | Participants66 | EvidenceC | Evaluated |
| Rank08 | ModelSA | Score92.1% | Percentile89.2% | Participants66 | EvidenceC | Evaluated |
| Rank09 | ModelAN | Score92.0% | Percentile87.7% | Participants66 | EvidenceC | Evaluated |
| Rank10 | ModelMA | Score92.0% | Percentile86.2% | Participants66 | EvidenceC | Evaluated |
| Rank11 | ModelAC | Score91.5% | Percentile84.6% | Participants66 | EvidenceC | Evaluated |
| Rank12 | ModelOP | Score90.2% | Percentile83.1% | Participants66 | EvidenceC | Evaluated |
| Rank13 | ModelIB | Score89.7% | Percentile81.5% | Participants66 | EvidenceC | Evaluated |
| Rank14 | ModelIB | Score89.7% | Percentile80.0% | Participants66 | EvidenceC | Evaluated |
| Rank15 | ModelGO | Score89.6% | Percentile78.5% | Participants66 | EvidenceC | Evaluated |
| Rank16 | ModelDE | Score89.0% | Percentile76.9% | Participants66 | EvidenceC | Evaluated |
| Rank17 | ModelME | Score89.0% | Percentile75.4% | Participants66 | EvidenceC | Evaluated |
| Rank18 | ModelAM Nova ProAmazon | Score89.0% | Percentile73.8% | Participants66 | EvidenceC | Evaluated |
| Rank19 | ModelME | Score88.4% | Percentile72.3% | Participants66 | EvidenceC | Evaluated |
| Rank20 | ModelMA | Score88.4% | Percentile70.8% | Participants66 | EvidenceC | Evaluated |
| Rank21 | ModelXA | Score88.4% | Percentile69.2% | Participants66 | EvidenceC | Evaluated |
| Rank22 | ModelME | Score88.4% | Percentile67.7% | Participants66 | EvidenceC | Evaluated |
| Rank23 | ModelAC | Score88.4% | Percentile66.2% | Participants66 | EvidenceC | Evaluated |
| Rank24 | ModelAC | Score88.4% | Percentile64.6% | Participants66 | EvidenceC | Evaluated |
| Rank25 | ModelAN | Score88.1% | Percentile63.1% | Participants66 | EvidenceC | Evaluated |
| Rank26 | ModelOP | Score88.1% | Percentile61.5% | Participants66 | EvidenceC | Evaluated |
| Rank27 | ModelOP | Score88.0% | Percentile60.0% | Participants66 | EvidenceC | Evaluated |
| Rank28 | ModelGO | Score87.8% | Percentile58.5% | Participants66 | EvidenceC | Evaluated |
| Rank29 | ModelOP | Score87.2% | Percentile56.9% | Participants66 | EvidenceC | Evaluated |
| Rank30 | ModelOP | Score87.1% | Percentile55.4% | Participants66 | EvidenceC | Evaluated |
| Rank31 | ModelAC | Score86.6% | Percentile53.9% | Participants66 | EvidenceC | Evaluated |
| Rank32 | ModelAC | Score86.0% | Percentile52.3% | Participants66 | EvidenceC | Evaluated |
| Rank33 | ModelXA | Score85.7% | Percentile50.8% | Participants66 | EvidenceC | Evaluated |
| Rank34 | ModelGO | Score85.4% | Percentile49.2% | Participants66 | EvidenceC | Evaluated |
| Rank35 | ModelAMNova LiteAmazon | Score85.4% | Percentile47.7% | Participants66 | EvidenceC | Evaluated |
| Rank36 | ModelAN | Score84.9% | Percentile46.1% | Participants66 | EvidenceC | Evaluated |
| Rank37 | ModelMA | Score84.8% | Percentile44.6% | Participants66 | EvidenceC | Evaluated |
| Rank38 | ModelAC | Score84.8% | Percentile43.1% | Participants66 | EvidenceC | Evaluated |
| Rank39 | ModelGO | Score84.1% | Percentile41.5% | Participants66 | EvidenceC | Evaluated |
| Rank40 | ModelAC | Score83.5% | Percentile40.0% | Participants66 | EvidenceC | Evaluated |
| Rank41 | ModelMI | Score82.6% | Percentile38.5% | Participants66 | EvidenceC | Evaluated |
| Rank42 | ModelIB | Score82.4% | Percentile36.9% | Participants66 | EvidenceC | Evaluated |
| Rank43 | ModelMA | Score81.1% | Percentile35.4% | Participants66 | EvidenceC | Evaluated |
| Rank44 | ModelAMNova MicroAmazon | Score81.1% | Percentile33.9% | Participants66 | EvidenceC | Evaluated |
| Rank45 | ModelME | Score80.5% | Percentile32.3% | Participants66 | EvidenceC | Evaluated |
| Rank46 | ModelAC | Score79.9% | Percentile30.8% | Participants66 | EvidenceC | Evaluated |
| Rank47 | ModelAC | Score78.7% | Percentile29.2% | Participants66 | EvidenceC | Evaluated |
| Rank48 | ModelAN | Score75.9% | Percentile27.7% | Participants66 | EvidenceC | Evaluated |
| Rank49 | ModelGO | Score75.0% | Percentile26.1% | Participants66 | EvidenceC | Evaluated |
| Rank50 | ModelGO | Score75.0% | Percentile24.6% | Participants66 | EvidenceC | Evaluated |
| Rank51 | ModelGO | Score74.3% | Percentile23.1% | Participants66 | EvidenceC | Evaluated |
| Rank52 | ModelXA | Score74.1% | Percentile21.5% | Participants66 | EvidenceC | Evaluated |
| Rank53 | ModelAN | Score73.0% | Percentile20.0% | Participants66 | EvidenceC | Evaluated |
| Rank54 | ModelME | Score72.6% | Percentile18.5% | Participants66 | EvidenceC | Evaluated |
| Rank55 | ModelMA | Score72.0% | Percentile16.9% | Participants66 | EvidenceC | Evaluated |
| Rank56 | ModelGO | Score71.3% | Percentile15.4% | Participants66 | EvidenceC | Evaluated |
| Rank57 | ModelMI | Score70.7% | Percentile13.8% | Participants66 | EvidenceC | Evaluated |
| Rank58 | ModelOP | Score68.0% | Percentile12.3% | Participants66 | EvidenceB | Evaluated |
| Rank59 | ModelOP | Score67.0% | Percentile10.8% | Participants66 | EvidenceC | Evaluated |
| Rank60 | ModelGO | Score66.5% | Percentile9.2% | Participants66 | EvidenceC | Evaluated |
| Rank61 | ModelGO | Score66.5% | Percentile7.7% | Participants66 | EvidenceC | Evaluated |
| Rank62 | ModelMI | Score62.8% | Percentile6.2% | Participants66 | EvidenceC | Evaluated |
| Rank63 | ModelGO | Score51.8% | Percentile4.6% | Participants66 | EvidenceC | Evaluated |
| Rank64 | ModelGO | Score41.5% | Percentile3.1% | Participants66 | EvidenceC | Evaluated |
| Rank65 | ModelGO | Score40.2% | Percentile1.5% | Participants66 | EvidenceC | Evaluated |
| Rank66 | ModelMA | Score34.8% | Percentile0.0% | 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.
Ranking basisThis humaneval AI model leaderboard uses descending score in the benchmark's original unit. The leaderboard ranking keeps matched price and speed data separate from benchmark evidence.
- 01OPMiniCPM-SALAOpenBMB
- Score
- 95.1%
Strengths
- Ranks #1 of 66 compared models
- 100th percentile on this benchmark
- C evidence result
Considerations
- This result measures HumanEval, not total model capability
- 02MAKimi K2 0905Moonshot AI
- Score
- 94.5%
- Price
- $0.60 input / $2.5 output per 1M tokens
- Speed
- Up to 45 tok/s via Novita
Strengths
- Ranks #2 of 66 compared models
- 98th percentile on this benchmark
- C evidence result
Considerations
- This result measures HumanEval, not total model capability
- 03ANClaude 3.5 SonnetAnthropic
- Score
- 93.7%
- Speed
- Up to 101 tok/s via Bedrock
Strengths
- Ranks #3 of 66 compared models
- 97th percentile on this benchmark
- C evidence result
Considerations
- This result measures HumanEval, not total model capability
- 04OPGPT-5OpenAI
- Score
- 93.4%
- Price
- $1.3 input / $10 output per 1M tokens
- Speed
- Up to 100 tok/s via OpenAI
Strengths
- Ranks #4 of 66 compared models
- 95th percentile on this benchmark
- C evidence result
Considerations
- This result measures HumanEval, not total model capability
- 05MAKimi K2 InstructMoonshot AI
- Score
- 93.3%
- Price
- $0.60 input / $2.5 output per 1M tokens
- Speed
- Up to 45 tok/s via Novita
Strengths
- Ranks #5 of 66 compared models
- 94th percentile on this benchmark
- C evidence result
Considerations
- This result measures HumanEval, not total model capability
Selection summary
Best AI Models for HumanEval
MiniCPM-SALA currently leads HumanEval with 95.1%. It is the top model on this specific benchmark, while the best LLM for the broader task should also be checked against other benchmarks, price and runtime.
Use this leaderboard with the supporting benchmark results and coverage details above. A leaderboard position summarizes the selected ranking signal; it does not replace workload-specific testing.
What is HumanEval?
What HumanEval measures and how its scores work.
A benchmark that measures functional correctness for synthesizing programs from docstrings, consisting of 164 original programming problems assessing 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
- Yes
- Evaluation key
- humaneval|llm-stats-current
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 95.1%.
What does HumanEval measure?
A benchmark that measures functional correctness for synthesizing programs from docstrings, consisting of 164 original programming problems assessing language comprehension, algorithms, and simple mathematics
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
66 model results are currently shown.
Does this benchmark affect the overall score?
Yes. This benchmark can contribute to the current LLMBoard capability score.