Coding benchmark
Vibe Code Bench leaderboard
Vibe Code Bench v1.1. Every model the catalog carries a published Vibe Code Bench value for, ranked by that value.
Vals.ai benchmark for evaluating whether models can build complete web applications from natural language specifications in a production-like development environment.
Vibe Code Bench ranking
41 models with a published Vibe Code Bench value, ordered by that value, highest first. Models the source has not scored on this benchmark are not listed — they are unmeasured, not last.
| Rank | Model | Provider | Full-stack app implementation benchmark |
|---|---|---|---|
| 1 |  Claude Opus 4.7 | Anthropic | 71.00 |
| 2 |  GPT-5.5 | OpenAI | 69.85 |
| 3 | GPT-5.4 | OpenAI | 67.42 |
| 4 | GPT-5.3 Codex | OpenAI | 61.77 |
| 5 | Claude Opus 4.6 | Anthropic | 57.57 |
| 6 | GPT-5.2 | OpenAI | 53.50 |
| 7 | Claude Opus 4.6 (Adaptive) | Anthropic | 53.50 |
| 8 | Claude Sonnet 4.6 | Anthropic | 51.48 |
| 9 |  DeepSeek V4 Pro 0813 | DeepSeek | 49.93 |
| 10 |  Gemini 3.5 Flash | 48.68 | |
| 11 | GPT-5.4 mini | OpenAI | 47.97 |
| 12 | GPT-5.2-Codex | OpenAI | 37.91 |
| 13 |  Kimi K2.6 | Moonshot AI | 37.89 |
| 14 | Gemini 3.1 Pro | 32.03 | |
| 15 |  GLM-5.1 | Z.AI | 31.46 |
| 16 |  MiniMax M2.7 | MiniMax | 27.04 |
| 17 | GPT-5.4 nano | OpenAI | 26.10 |
| 18 |  Qwen3.6 Plus | Alibaba | 25.56 |
| 19 | GPT-5.1 | OpenAI | 24.61 |
| 20 | GLM-5 (Reasoning) | Z.AI | 23.36 |
| 21 | Claude Sonnet 4.5 Thinking | Anthropic | 22.62 |
| 22 | GPT-5.1-Codex-Max | OpenAI | 22.17 |
| 23 | Claude Opus 4.5 Thinking | Anthropic | 20.63 |
| 24 | Gemini 3 Flash | 20.20 | |
| 25 | GPT-5 (high) | OpenAI | 20.09 |
| 26 |  Muse Spark | Meta | 19.67 |
| 27 | Kimi K2.5 (Reasoning) | Moonshot AI | 17.54 |
| 28 | Qwen3.5 Plus | Alibaba | 15.74 |
| 29 | MiniMax M2.5 | MiniMax | 14.85 |
| 30 | Gemini 3 Pro | 14.30 | |
| 31 | GPT-5 mini | OpenAI | 14.17 |
| 32 | GPT-5.1-Codex | OpenAI | 13.12 |
| 33 | Claude Haiku 4.5 Thinking | Anthropic | 11.39 |
| 34 | DeepSeek V3.2 (Thinking) | DeepSeek | 5.11 |
| 35 |  Grok 4.20 | xAI | 4.06 |
| 36 | Qwen3 Max | Alibaba | 3.51 |
| 37 | GLM-4.6 | Z.AI | 3.09 |
| 38 | Grok 4.1 Fast (Reasoning) | xAI | 1.20 |
| 39 | Gemini 2.5 Pro | 0.40 | |
| 40 | Gemini 3.1 Flash-Lite | 0.00 | |
| 40 | Grok 4 Fast (Reasoning) | xAI | 0.00 |
Evidence key: Observed
Rows are ordered by the value Vibe Code Bench: Evaluating AI Models on End-to-End Web Application Development published, highest first. The source does not state whether a higher value is the better result, so this page does not either: for a benchmark that measures a rate of failure, read the table from the bottom.