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

Metric: Instruction-Following Rate (%; 1,286 ICC tasks: passes the unit tests and a Claude3.5-Sonnet judge finds the instructed implementation). Source: arxiv.org. Saturation forecast: Estimated already saturated. 42 models tracked.

Top models

#ModelScore
1O1 (2024-12-17)62.9
2Claude 3.5 Sonnet (20241022)60.9
3GPT-4o (2024-11-20)59.6
4Gemini 2.0 Flash59.5
5GPT-4o (2024-08-06)58.6
6O1 Preview57.7
7O1 Mini55
8DeepSeek V3 (0324)53
9DeepSeek V347.3
10GPT-4o Mini (2024-07-18)42.2
11GPT-4o (2024-05-13)41.3
12Claude 3.5 Haiku40.7
13Qwen 2.5 Coder 32B38.7
14Qwen 2.5 Coder 14B36.2
15Codestral-22B-v0.134.1

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