benchmark evidence
Aider Coding Benchmark
Real-world coding task completion via Aider edit benchmark.
winner on Aider Coding Benchmark
direct benchmark result, not a broad vertical composite | source row dated 2026-05-08
scored on 2026-05-08
latest mapped results | top 20
| # | Model | Score | Evidence | Tested |
|---|---|---|---|---|
| 1 | Google: Gemini 2.5 Pro | 60.9 | model-only independent_benchmark | 2026-05-08 |
| 2 | OpenAI: GPT-5 | 58.6 | model-only independent_benchmark | 2026-05-08 |
| 3 | OpenAI: o1 | 57.9 | model-only independent_benchmark | 2026-05-08 |
| 4 | Google: Gemini 2.0 Flash | 56.4 | model-only independent_benchmark | 2026-05-08 |
| 5 | Google: Gemini 2.0 Flash Lite | 56.4 | model-only independent_benchmark | 2026-05-08 |
| 6 | Qwen2.5 Coder 32B Instruct | 49.6 | model-only independent_benchmark | 2026-05-08 |
| 7 | Amazon: Nova Pro 1.0 | 44.4 | model-only independent_benchmark | 2026-05-08 |
| 8 | Meta: Llama 3.3 70B Instruct | 42.1 | model-only independent_benchmark | 2026-05-08 |
what this result means
Real-world coding task completion via Aider edit benchmark.
This benchmark contributes direct public evidence. Read its scope before generalizing the result.
A win here is a win on Aider Coding Benchmark. Broad task pages require independent corroboration before naming a general winner.
source record
category: coding
metric: pass@1
matched models: 8
latest source date: 2026-05-08
direction: higher is better