Event ·Multi-source

reactbench.com: 全新 React 编码基准测试发布

reactbench.com 是专门针对 React 编码的真实基准测试，包含 51 个真实任务，均来自开源项目实际 PR 和 Issue。测试结果显示 GPT-5.6 Sol xHigh 综合表现最强，成本仅 1.43 美元。GPT-5.6 Terra Medium 性价比最佳，成本 0.53 美元，而 Claude Fable 5 成本高达 9.05 美元，约为 Sol 的 6-7 倍。

Current read

百万.js 作者搞了个 React 编码实测，51个真实任务，GPT 5.6 最便宜最强，Claude 太贵了。

11 sources51° AI heatLast updated 2026/7/16 03:13:52

Evidence

11 sources

Primary sourceViking

2026/7/16 03:13:52View source

Related source 1Amjad Masad

2026/7/17 00:08:43View source

Related source 2Together AI

2026/7/18 01:38:10View source

Related source 3lmarena.ai

2026/7/15 16:37:10View source

Related source 4Fireworks AI

2026/7/15 22:35:30View source

Related source 5orange.ai

2026/7/17 02:23:30View source

Related source 6AI Will

2026/7/17 08:04:14View source

Related source 7Jerry Liu

2026/7/17 15:35:55View source

Related source 8Thomas Wolf

2026/7/17 18:37:40View source

Related source 9Decoder

2026/7/16 19:49:39View source

Related source 10arXiv cs.AI

2026/7/17 17:46:23View source

Conflict check

Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.

Open the story