benchmark evidence
WebDev Arena
Arena.ai WebDev / Code Arena frontier coding preference leaderboard.
winner on WebDev Arena
direct benchmark result, not a broad vertical composite | source row dated 2026-09-01
scored on 2026-09-01
latest mapped results | top 20
| # | Model | Score | Evidence | Tested |
|---|---|---|---|---|
| 1 | Anthropic: Claude Opus 4.7 | 100.0 | model-only independent_benchmark | 2026-09-01 |
| 2 | Anthropic: Claude Opus 4.6 | 100.0 | model-only independent_benchmark | 2026-09-01 |
| 3 | Anthropic: Claude Sonnet 4.6 | 100.0 | model-only independent_benchmark | 2026-09-01 |
| 4 | Anthropic: Claude Opus 4.5 | 92.0 | model-only independent_benchmark | 2026-09-01 |
| 5 | DeepSeek: DeepSeek V4 Pro 0423 | 86.4 | model-only independent_benchmark | 2026-09-01 |
| 6 | Z.ai: GLM 5 | 83.9 | model-only independent_benchmark | 2026-09-01 |
| 7 | OpenAI: GPT-5.2 | 79.1 | model-only independent_benchmark | 2026-09-01 |
| 8 | OpenAI: GPT-5.4 Mini | 74.3 | model-only independent_benchmark | 2026-09-01 |
| 9 | OpenAI: GPT-5.4 | 72.2 | model-only independent_benchmark | 2026-09-01 |
| 10 | Anthropic: Claude Sonnet 4.5 | 71.4 | model-only independent_benchmark | 2026-09-01 |
| 11 | OpenAI: GPT-5.1 | 60.2 | model-only independent_benchmark | 2026-09-01 |
| 12 | Anthropic: Claude Haiku 4.5 | 57.2 | model-only independent_benchmark | 2026-09-01 |
| 13 | DeepSeek: DeepSeek V3.2 | 56.2 | model-only independent_benchmark | 2026-09-01 |
| 14 | Google: Gemini 2.5 Pro | 31.4 | model-only independent_benchmark | 2026-09-01 |
what this result means
Arena.ai WebDev / Code Arena frontier coding preference leaderboard.
This benchmark contributes direct public evidence. Read its scope before generalizing the result.
A win here is a win on WebDev Arena. Broad task pages require independent corroboration before naming a general winner.
source record
category: coding
metric: accuracy
matched models: 14
latest source date: 2026-09-01
direction: higher is better