C3-Bench (Controllable Code Completion) - Scale Control: leaderboard

Metric: Instruction-Following Rate (%; 909 SCC tasks: AST node-type and line-count check of the requested completion scope). Source: arxiv.org. Saturation forecast: Estimated already saturated. 42 models tracked.

Top models

#ModelScore
1O1 (2024-12-17)59.6
2Claude 3.5 Sonnet (20241022)50.8
3O1 Preview48.9
4O1 Mini44.7
5Claude 3.5 Haiku26.4
6GPT-4o (2024-08-06)24.1
7GPT-4o (2024-11-20)24.1
8GPT-4o Mini (2024-07-18)22.5
9GPT-4o (2024-05-13)20.8
10DeepSeek V320.2
11DeepSeek V3 (0324)20.2
12Qwen 2.5 Coder 32B Instruct16.9
13Qwen 2.5 Coder 14B Instruct13.5
14Qwen 2.5 72B Instruct9.4
15Qwen 2.5 Coder 14B8.7

Interactive version: theaggregate.ai/benchmark?slug=c3-bench-controllable-code-completion-scale-control · How It Works · Data refreshed daily, snapshot 2026-09-25.