- Home
- Benchmarks
- Aider Polyglot
Code generation · editing · in the overall grade, weight 3
Aider Polyglot
Hard Exercism exercises in C++, Go, Java, JavaScript, Python and Rust, solved by editing files through the Aider coding assistant.
- Models
- 8
- By makers
- 3
- Latest
- Sep 9
| # | Model | Bar | Score | Source |
|---|---|---|---|---|
| 1 | DeepSeek V3.1DeepSeek · $0.42/1M · Coming soon | 76.3%vAug 21, 2025 | MakerAug 21, 2025 | |
| 2 | DeepSeek R1 0528DeepSeek · $0.91/1M · Coming soon | 71.4%Jun 6, 2025 | LeaderboardJun 6, 2025 | |
| 3 | Qwen3 235B A22B Instruct 2507Qwen · $0.20/1M · Coming soon | 57.3%vJul 21, 2025 | MakerJul 21, 2025 | |
| 4 | Qwen3 Next 80B A3B InstructQwen · $0.34/1M · Coming soon | 49.8%vSep 9, 2025 | MakerSep 9, 2025 | |
| 5 | DeepSeek V3DeepSeek · $0.46/1M · Coming soon | 48.4%Dec 25, 2024 | LeaderboardDec 25, 2024 | |
| 6 | GPT OSS 120BOpenAI · $0.070/1M · Coming soon | 41.8%Aug 6, 2025 | LeaderboardAug 6, 2025 | |
| 7 | Qwen3 32BQwen · $0.13/1M · Coming soon | 40%May 8, 2025 | LeaderboardMay 8, 2025 | |
| 8 | Gemma 3 27B ITGoogle · $0.10/1M · Coming soon | 4.9%Mar 15, 2025 | LeaderboardMar 15, 2025 |
% of exercises solved. v marks a score the model’s maker reported. Prices are FullsendAI’s per million tokens, blended 3 input : 1 output.
About the data
From Aider (aider.chat). Aider is Apache-2.0 licensed, its leaderboard data included; reuse with attribution to aider.chat. No longer actively updated (last entry Oct 2025).