Models72°

GPT-6.2 Astra (Max) 登上 Agent Arena 榜第6名,编码涨 50 分、聊天掉 54 分

Why we picked it

OpenAI 新版 GPT-6.2 Astra 编码分暴涨 50 分冲到第 3,但聊天分掉了 54 分。写代码的选它没毛病,日常问答类任务还得挑别的模型,选型前看清分项数据。

OpenAI 的 GPT-6.2 Astra (Max) 在 Agent Arena Index 总榜排名第 6,比 GPT-6 Astra (Max) 高一位。分项分数显示编码从 1,288 升至 1,338(+50),排名从第 8 升到第 3;Work 从 1,267 升至 1,280(+13)。

Body

Chat 方向则从 1,265 降至 1,211(-54),排名从第 8 掉到第 11。也就是说,新模型在研究、写作、辅导等日常任务上变弱,编码能力增强,两个变化在总分上几乎互相抵消。

对开发者来说,如果工作负载以代码为主,GPT-6.2 Astra 是更优选择;通用对话场景可能仍需保留 GPT-6 Astra 或其他 Chat 排名更高的模型。

Original · lmarena.ai

GPT-6.2 Astra (Max) by @OpenAI ranks #6 on the Agent Arena Index, one place above GPT-6 Astra (Max). However, these models excel in very different capabilities. Let’s dive in by index. GPT-6.2 Astra vs GPT-6 Astra across key Agent Arena Indexes: - Coding: 1,338 vs 1,288 (+50) - Work: 1,280 vs 1,267 (+13) - Chat: 1,211 vs 1,265 (-54) This shows the new model’s gain is in coding, where GPT-6.2 Astra climbs from #8 to #3 . On Chat, the everyday tasks such as research, tutoring or writing, it drops from #8 to #11 . The two shifts nearly cancel out in the overall score.📊 15 ⚡

Industry Coordinates & Source Verification

This report is categorized under AITOP's "Models" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include GPT-6.2, OpenAI, Agent Arena, coding models.

Source attribution: curated from lmarena.ai (@lmarena_ai).