VibeCodingBench — leaderboard
Production-oriented coding benchmark evaluating AI coding agents across functional correctness, visual fidelity, code quality, security, cost, and speed on representative developer tasks.
Metric: Average Score. Source: github.com. 15 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus (Claude Opus 4.5) | 89.15 |
| 2 | Claude Haiku (Claude Haiku 4.5) | 88.97 |
| 3 | Grok (Grok 4 Fast) | 88.8 |
| 4 | OpenAI (OpenAI GPT-5.2) | 88.75 |
| 5 | Qwen (Qwen3 Max) | 88.6 |
| 6 | Claude (Claude Sonnet 4.5) | 88.56 |
| 7 | GLM-4 (GLM 4-Plus) | 88.2 |
| 8 | DeepSeek (DeepSeek v3.2) | 88.19 |
| 9 | Grok (Grok 4) | 88 |
| 10 | MiniMax (MiniMax M2.1) | 87.42 |
Interactive version: theaggregate.ai/benchmark?slug=vibecodingbench · How the rankings work · Data refreshed daily, snapshot 2026-07-22.