Coding · Benchmark profile
Vals.ai benchmark for evaluating whether models can build complete web applications from natural language specifications in a production-like development environment.
Data verified 27 Jul 2026 · Methodology 1.8.0
Benchmark score on Vibe Code Bench v1.1
Claude Opus 4.7 leads at 71.0%, followed by GPT-5.5 (69.8%) and GPT-5.4 (67.4%).
10 ranked models · higher is better · labels show rank and score
| Model | Rank | Provider | Score |
|---|---|---|---|
| Claude Opus 4.7 | #1 | Anthropic | 71.0% |
| GPT-5.5 | #2 | OpenAI | 69.8% |
| GPT-5.4 | #3 | OpenAI | 67.4% |
| GPT-5.3-Codex | #4 | OpenAI | 61.8% |
| Claude Opus 4.6 | #5 | Anthropic | 57.6% |
| GPT-5.2 | #6 | OpenAI | 53.5% |
| Claude Opus 4.6 (Adaptive) | #7 | Anthropic | 53.5% |
| Claude Sonnet 4.6 | #8 | Anthropic | 51.5% |
| DeepSeek V4 Pro (Max) | #9 | DeepSeek | 49.9% |
| Gemini 3.5 Flash | #10 | 48.7% |
One best score per model · higher is better
| Rank | Model | Provider | License | Evidence use | Score |
|---|---|---|---|---|---|
| #1 | Claude Opus 4.7 claude-opus-4-7 | Anthropic | closed | Reference only | 71.0% |
| #2 | GPT-5.5 gpt-5.5 | OpenAI | closed | Estimated reference | 69.8% |
| #3 | GPT-5.4 gpt-5.4 | OpenAI | closed | Reference only | 67.4% |
| #4 | GPT-5.3-Codex gpt-5.3-codex | OpenAI | closed | Reference only | 61.8% |
| #5 | Claude Opus 4.6 claude-opus-4-6 | Anthropic | closed | Reference only | 57.6% |
| #6 | GPT-5.2 gpt-5.2 | OpenAI | closed | Estimated reference | 53.5% |
| #7 | Claude Opus 4.6 (Adaptive) claude-opus-4-6-thinking | Anthropic | closed | Estimated reference | 53.5% |
| #8 | Claude Sonnet 4.6 claude-sonnet-4-6 | Anthropic | closed | Reference only | 51.5% |
| #9 | DeepSeek V4 Pro (Max) deepseek-v4-pro-max | DeepSeek | open | Estimated reference | 49.9% |
| #10 | Gemini 3.5 Flash gemini-3.5-flash | closed | Estimated reference | 48.7% | |
| #11 | GPT-5.4 mini gpt-5.4-mini | OpenAI | closed | Estimated reference | 48.0% |
| #12 | GPT-5.2 Codex gpt-5.2-codex | OpenAI | closed | Reference only | 37.9% |
| #13 | Kimi K2.6 kimi-k2.6 | Moonshot AI | open | Estimated reference | 37.9% |
| #14 | Gemini 3.1 Pro gemini-3-1-pro | closed | Estimated reference | 32.0% | |
| #15 | GLM-5.1 glm-5-1 | Z.AI | open | Reference only | 31.5% |
| #16 | MiniMax M2.7 minimax-m2-7 | MiniMax | open | Reference only | 27.0% |
| #17 | GPT-5.4 nano gpt.5.4.nano | OpenAI | closed | Reference only | 26.1% |
| #18 | Qwen3.6 Plus qwen3-6-plus | Alibaba Cloud | closed | Reference only | 25.6% |
| #19 | GPT-5.1 gpt.5.1 | OpenAI | closed | Estimated reference | 24.6% |
| #20 | GLM-5 (Reasoning) glm-5-reasoning | Z.AI | open | Estimated reference | 23.4% |
| #21 | Claude Sonnet 4.5 Thinking claude-sonnet-4-5-thinking | Anthropic | closed | Estimated reference | 22.6% |
| #22 | GPT-5.1-Codex-Max gpt-5-1-codex-max | OpenAI | closed | Estimated reference | 22.2% |
| #23 | Claude Opus 4.5 Thinking claude-opus-4-5-thinking | Anthropic | closed | Estimated reference | 20.6% |
| #24 | Gemini 3 Flash gemini-3-flash | closed | Reference only | 20.2% | |
| #25 | GPT-5 (high) gpt-5-high | OpenAI | closed | Estimated reference | 20.1% |
| #26 | Muse Spark muse-spark | Meta | closed | Reference only | 19.7% |
| #27 | Kimi K2.5 (Reasoning) kimi-k2-5-reasoning | Moonshot AI | closed | Estimated reference | 17.5% |
| #28 | Qwen3.5 Plus qwen3-5-plus | Alibaba Cloud | closed | Estimated reference | 15.7% |
| #29 | MiniMax M2.5 minimax-m2.5 | MiniMax | open | Reference only | 14.9% |
| #30 | Gemini 3 Pro gemini-3-pro | closed | Reference only | 14.3% | |
| #31 | GPT-5 mini gpt-5-mini | OpenAI | closed | Reference only | 14.2% |
| #32 | GPT-5.1-Codex gpt-5-1-codex | OpenAI | closed | Estimated reference | 13.1% |
| #33 | Claude Haiku 4.5 Thinking claude-haiku-4-5-thinking | Anthropic | closed | Estimated reference | 11.4% |
| #34 | DeepSeek V3.2 (Thinking) deepseek-v3-2-thinking | DeepSeek | open | Estimated reference | 5.1% |
| #35 | Grok 4.20 grok-4.20 | xAI | closed | Estimated reference | 4.1% |
| #36 | Qwen3 Max qwen3-max | Alibaba Cloud | closed | Estimated reference | 3.5% |
| #37 | GLM-4.6 glm-4.6 | Z.AI | open | Reference only | 3.1% |
| #38 | Grok 4.1 Fast (Reasoning) grok-4-1-fast-reasoning | xAI | closed | Reference only | 1.2% |
| #39 | Gemini 2.5 Pro gemini-2.5-pro | closed | Reference only | 0.4% | |
| #40 | Gemini 3.1 Flash-Lite gemini-3.1-flash-lite | closed | Reference only | 0% | |
| #41 | Grok 4 Fast (Reasoning) grok-4-fast-reasoning | xAI | closed | Reference only | 0% |
The top of this snapshot is led by Claude Opus 4.7 at 71.0%; third place is 3.6 points behind. The top-10 spread is 22.3 points.
About Vibe Code Bench v1.1
Vals.ai benchmark for evaluating whether models can build complete web applications from natural language specifications in a production-like development environment. Results stay tied to the exact model variant and evaluation system. Multiple systems for the same model use the best published score on this page; overall Lumina scoring uses the median of ranking-eligible rows.
Open benchmark source ↗FAQ
Vals.ai benchmark for evaluating whether models can build complete web applications from natural language specifications in a production-like development environment.
Claude Opus 4.7 by Anthropic currently leads with 71.0%.
41 models in the LuminaBench cohort have a qualifying score on this benchmark.
No. This benchmark is display-only and does not enter the overall Lumina composite.
Related
