AiderFrom the builder
o1-preview is SOTA on the aider leaderboard
Aider published first results for OpenAI's o1 models on its code editing benchmark, where o1-preview set a top score of 79.7% using the whole-file edit format. With the more practical diff format it scored 75.2%, which Aider says likely puts it between Sonnet and GPT-4o in everyday use at much higher cost. The o1-mini model, priced close to GPT-4o and Claude 3.5 Sonnet, scored below both, and aider v0.57.0 supports both models.
The o1-preview model had trouble conforming to aider’s diff edit format.
Saved . A shortened excerpt from the official blog; the original publisher retains authorship.