general benchmark
Aider-Polyglot
A coding benchmark that evaluates LLMs on 225 challenging Exercism programming exercises across C++, Go, Java, JavaScript, Python, and Rust. Models receive two attempts to solve each problem, with test error feedback provided after the first attempt if it fails. The benchmark measures both initial problem-solving ability and capacity to edit code based on error feedback, providing an end-to-end evaluation of code generation and editing capabilities across multiple programming languages.
Updated Aug 7, 2026
Published models22
Registry coverage22
MetricScore
EvidenceB
Aider-Polyglot leaderboard
Sorted by the source-provided rank. Higher score is better according to the registry.
| 1 | OP | 88.0% | 100.0% | 22 | C | |
| 2 | GO | 82.2% | 95.2% | 22 | C | |
| 3 | OP | 81.3% | 90.5% | 22 | C | |
| 4 | GO | 76.5% | 85.7% | 22 | C | |
| 5 | DE | 74.5% | 81.0% | 22 | C | |
| 6 | DE | 71.6% | 76.2% | 22 | C | |
| 7 | OP | 68.9% | 71.4% | 22 | C | |
| 8 | DE | 68.4% | 66.7% | 22 | C | |
| 9 | OP | 66.7% | 61.9% | 22 | C | |
| 10 | GO | 61.9% | 57.1% | 22 | C | |
| 11 | AC | 61.8% | 52.4% | 22 | C | |
| 12 | MA | 60.0% | 47.6% | 22 | C | |
| 13 | MA | 60.0% | 42.9% | 22 | C | |
| 14 | AC | 57.3% | 38.1% | 22 | C | |
| 15 | OP | 51.6% | 33.3% | 22 | C | |
| 16 | AC | 49.8% | 28.6% | 22 | C | |
| 17 | DE | 49.6% | 23.8% | 22 | C | |
| 18 | MA | 47.1% | 19.1% | 22 | C | |
| 19 | OP | 34.7% | 14.3% | 22 | C | |
| 20 | OP | 30.7% | 9.5% | 22 | C | |
| 21 | GO | 26.7% | 4.8% | 22 | C | |
| 22 | OP | 9.8% | 0.0% | 22 | C |
Score distribution
Top published rows on the benchmark's original scale.
Aider-Polyglot highlights
The top published results on this benchmark's own scale.
What is Aider-Polyglot?
Definition and scoring fields from the benchmark registry.
A coding benchmark that evaluates LLMs on 225 challenging Exercism programming exercises across C++, Go, Java, JavaScript, Python, and Rust. Models receive two attempts to solve each problem, with test error feedback provided after the first attempt if it fails. The benchmark measures both initial problem-solving ability and capacity to edit code based on error feedback, providing an end-to-end evaluation of code generation and editing capabilities across multiple programming languages.
Scores are shown in ratio. The current registry marks this benchmark as not independently verified with evidence level B.
- Family
- Aider-Polyglot
- Modality
- text
- Primary category
- general
- Score direction
- higher
- LLMBoard eligible
- Yes
- Evaluation key
- aider-polyglot|llm-stats-current
Source-native results are preserved. Eligibility for the overall LLMBoard score is a separate policy decision.
FAQ
Common questions about Aider-Polyglot.
Which model scores highest on Aider-Polyglot?
GPT-5 is currently ranked first with 88.0%.
What does Aider-Polyglot measure?
A coding benchmark that evaluates LLMs on 225 challenging Exercism programming exercises across C++, Go, Java, JavaScript, Python, and Rust. Models receive two attempts to solve each problem, with test error feedback provided after the first attempt if it fails. The benchmark measures both initial problem-solving ability and capacity to edit code based on error feedback, providing an end-to-end evaluation of code generation and editing capabilities across multiple programming languages.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
22 unique published model results are currently shown.
Does this benchmark affect the overall score?
This benchmark is marked as eligible for the current LLMBoard capability methodology.