225 hard Exercism exercises across six languages, scoring whether a model can write and then fix its own code from failing test output.
unassessed
| Category | coding |
|---|---|
| Subcategory | multi-language code editing |
| Page status | active |
| Metric | pass rate (2 attempts) |
| Direction | higher_is_better |
| Unit | % |
| Dataset size | 225 |
| Dataset licence | Exercise content is Exercism's, used under Exercism's per-track open-source licences |
| Publisher | Aider (Paul Gauthier) |
The model is given an Exercism programming exercise description and must produce a working solution, expressed as an edit to a starting file, in one of six languages: C++, Go, Java, JavaScript, Python or Rust. Unlike a single-shot coding benchmark, it also measures whether the model can act on its own failing test output to fix a first attempt.
Agentic code editing inside the aider CLI tool: the model receives the exercise prompt and must emit an edit in one of aider's supported edit formats, which aider applies to a real file and tests with the language's unit test suite.
| Model | Provider | Score | Card as of |
|---|---|---|---|
| Gemini 2.5 Pro Preview 06-05 | Google DeepMind | 83.1 | 2026-04 |
| Claude Opus 4 | Anthropic | 82.1 | 2026-04 |
| Claude Opus 4.6 | Anthropic | 82.1 | 2026-04 |
| o3 | OpenAI | 80.5 | 2026-04 |
| o3-deep-research | OpenAI | 80.5 | 2026-04 |
| o3-mini | OpenAI | 80.5 | 2026-04 |
| o3-pro | OpenAI | 80.5 | 2026-04 |
| Claude Opus 4.1 | Anthropic | 78.8 | 2026-04 |
| Claude Opus 4.1 (latest) | Anthropic | 78.8 | 2026-04 |
| Gemini 2.5 Pro Preview 05-06 | Google DeepMind | 76.9 | 2026-04 |
| Claude Sonnet 4 | Anthropic | 75.5 | 2026-04 |
| Claude Sonnet 4.5 | Anthropic | 75.5 | 2026-04 |
| Claude Sonnet 4.5 (latest) | Anthropic | 75.5 | 2026-04 |
| Gemini 2.5 Pro Preview TTS | Google DeepMind | 72.9 | 2026-04 |
| Qwen3-Coder 480B-A35B Instruct | Alibaba / Qwen Team | 72.5 | 2026-04 |
| GPT-5 | OpenAI | 72.1 | 2026-04 |
| GPT-5 Chat (latest) | OpenAI | 72.1 | 2026-04 |
| GPT-5 Mini | OpenAI | 72.1 | 2026-04 |
| GPT-5 Nano | OpenAI | 72.1 | 2026-04 |
| GPT-5 Pro | OpenAI | 72.1 | 2026-04 |
| GPT-5-Codex | OpenAI | 72.1 | 2026-04 |
| GPT-5.1 | OpenAI | 72.1 | 2026-04 |
| GPT-5.1 Chat | OpenAI | 72.1 | 2026-04 |
| GPT-5.1 Codex | OpenAI | 72.1 | 2026-04 |
| GPT-5.1 Codex Max | OpenAI | 72.1 | 2026-04 |
| GPT-5.1 Codex mini | OpenAI | 72.1 | 2026-04 |
| GPT-5.2 | OpenAI | 72.1 | 2026-04 |
| GPT-5.2 Chat | OpenAI | 72.1 | 2026-04 |
| GPT-5.2 Codex | OpenAI | 72.1 | 2026-04 |
| GPT-5.2 Pro | OpenAI | 72.1 | 2026-04 |
| GPT-5.3 Chat (latest) | OpenAI | 72.1 | 2026-04 |
| GPT-5.3 Codex | OpenAI | 72.1 | 2026-04 |
| GPT-5.3 Codex Spark | OpenAI | 72.1 | 2026-04 |
| GPT-5.4 | OpenAI | 72.1 | 2026-04 |
| GPT-5.4 mini | OpenAI | 72.1 | 2026-04 |
| GPT-5.4 nano | OpenAI | 72.1 | 2026-04 |
| GPT-5.4 Pro | OpenAI | 72.1 | 2026-04 |
| Claude Opus 4 (latest) | Anthropic | 72.0 | 2026-04 |
| o4-mini | OpenAI | 72.0 | 2026-04 |
| o4-mini-deep-research | OpenAI | 72.0 | 2026-04 |
| Gemini 2.5 Pro | Google DeepMind | 71.2 | 2026-04 |
| Grok 4 | xAI | 70.8 | 2026-04 |
| Grok 4 Fast | xAI | 70.8 | 2026-04 |
| Grok 4 Fast (Non-Reasoning) | xAI | 70.8 | 2026-04 |
| Grok 4.1 Fast | xAI | 70.8 | 2026-04 |
| Grok 4.1 Fast (Non-Reasoning) | xAI | 70.8 | 2026-04 |
| Grok 4.20 (Non-Reasoning) | xAI | 70.8 | 2026-04 |
| Grok 4.20 (Reasoning) | xAI | 70.8 | 2026-04 |
| Grok 4.20 Multi-Agent | xAI | 70.8 | 2026-04 |
| GPT-4.1 | OpenAI | 68.5 | 2026-04 |
| Qwen 3 235B Instruct | Cerebras | 68.2 | 2026-04 |
| Qwen3 235B-A22B | Alibaba / Qwen Team | 68.2 | 2026-04 |
| DeepSeek R1 | DeepSeek | 65.8 | 2026-04 |
| DeepSeek R1 0528 | DeepSeek | 65.8 | 2026-04 |
| DeepSeek R1 0528 NVFP4 v2 | NVIDIA | 65.8 | 2026-04 |
| DeepSeek Reasoner | DeepSeek | 65.8 | 2026-04 |
| Claude Sonnet 3.7 | Anthropic | 64.9 | 2026-04 |
| o1 | OpenAI | 61.7 | 2026-04 |
| o1-preview | OpenAI | 61.7 | 2026-04 |
| Claude Sonnet 4 (latest) | Anthropic | 61.3 | 2026-04 |
| Codestral (latest) | Mistral AI | 60.5 | 2026-04 |
| DeepSeek Chat | DeepSeek | 58.5 | 2026-04 |
| DeepSeek V3 | DeepSeek | 58.5 | 2026-04 |
| DeepSeek V3 0324 | DeepSeek | 58.5 | 2026-04 |
| DeepSeek V3.1 | DeepSeek | 58.5 | 2026-04 |
| DeepSeek V3.2 | DeepSeek | 58.5 | 2026-04 |
| DeepSeek V3.2 Exp | DeepSeek | 58.5 | 2026-04 |
| GPT-4o | OpenAI | 58.2 | 2026-04 |
| GPT-4o (2024-05-13) | OpenAI | 58.2 | 2026-04 |
| GPT-4o (2024-08-06) | OpenAI | 58.2 | 2026-04 |
| GPT-4o (2024-11-20) | OpenAI | 58.2 | 2026-04 |
| GPT-4o mini | OpenAI | 58.2 | 2026-04 |
| Gemini 2.5 Flash | Google DeepMind | 55.1 | 2026-04 |
| Gemini 2.5 Flash Image | Google DeepMind | 55.1 | 2026-04 |
| Gemini 2.5 Flash Image (Preview) | Google DeepMind | 55.1 | 2026-04 |
| Gemini 2.5 Flash Lite | Google DeepMind | 55.1 | 2026-04 |
| Gemini 2.5 Flash Lite Preview 06-17 | Google DeepMind | 55.1 | 2026-04 |
| Gemini 2.5 Flash Lite Preview 09-25 | Google DeepMind | 55.1 | 2026-04 |
| Gemini 2.5 Flash Preview 05-20 | Google DeepMind | 55.1 | 2026-04 |
| Gemini 2.5 Flash Preview 09-25 | Google DeepMind | 55.1 | 2026-04 |
| Gemini 2.5 Flash Preview TTS | Google DeepMind | 55.1 | 2026-04 |
| Gemma 4 31B | Google DeepMind | 54.8 | 2026-04 |
| gemma 4 31B it | Google DeepMind | 54.8 | 2026-04 |
| gemma 4 31B it GGUF | Unsloth | 54.8 | 2026-04 |
| Gemma 4 31B IT NVFP4 | NVIDIA | 54.8 | 2026-04 |
| Grok 3 | xAI | 53.3 | 2026-04 |
| Grok 3 Fast | xAI | 53.3 | 2026-04 |
| Grok 3 Fast Latest | xAI | 53.3 | 2026-04 |
| Grok 3 Latest | xAI | 53.3 | 2026-04 |
| Mistral Large (latest) | Mistral AI | 52.8 | 2026-04 |
| Mistral Large 2.1 | Mistral AI | 52.8 | 2026-04 |
| Mistral Large 3 | Mistral AI | 52.8 | 2026-04 |
| GPT-4 | OpenAI | 52.4 | 2026-04 |
| Gemma 4 26B | Google DeepMind | 52.1 | 2026-04 |
| Claude Sonnet 3.5 | Anthropic | 51.6 | 2026-04 |
| Claude Sonnet 3.5 v2 | Anthropic | 51.6 | 2026-04 |
| Grok 3 Mini | xAI | 49.3 | 2026-04 |
| Grok 3 Mini Fast | xAI | 49.3 | 2026-04 |
| Grok 3 Mini Fast Latest | xAI | 49.3 | 2026-04 |
| Grok 3 Mini Latest | xAI | 49.3 | 2026-04 |
| Gemini 2.5 Flash Preview 04-17 | Google DeepMind | 47.1 | 2026-04 |
| Llama 3.3 70B Instruct NVFP4 | NVIDIA | 45.2 | 2026-04 |
| Llama-3.3-70B-Instruct | Meta | 45.2 | 2026-04 |
| Qwen3 32B | Alibaba / Qwen Team | 40.0 | 2026-04 |
| Qwen3 32B AWQ | Alibaba / Qwen Team | 40.0 | 2026-04 |
| Qwen3 32B NVFP4 | NVIDIA | 40.0 | 2026-04 |
| o1-mini | OpenAI | 32.9 | 2026-04 |
| GPT-4.1 mini | OpenAI | 32.4 | 2026-04 |
| Claude Haiku 3.5 | Anthropic | 28.0 | 2026-04 |
| Claude Haiku 3.5 (latest) | Anthropic | 28.0 | 2026-04 |
| Qwen2.5 Coder 32B Instruct | Alibaba / Qwen Team | 16.4 | 2026-04 |
| Qwen2.5 Coder 32B Instruct AWQ | Alibaba / Qwen Team | 16.4 | 2026-04 |
| Llama 4 Maverick 17B 128E Instruct | Meta | 15.6 | 2026-04 |
| Llama-4-Maverick-17B-128E-Instruct-FP8 | Meta | 15.6 | 2026-04 |
| GPT-4.1 nano | OpenAI | 8.9 | 2026-04 |
| Gemma 3 27B | Google DeepMind | 4.9 | 2026-04 |