ModelsMulti-source74°

GoogleDeepMind 发布 Gemini 4 Neon,Agent Arena Index 排名第二

Just yesterday, @GoogleDeepMind released Gemini 4 Neon. Watch @petergostev’s first impressions as h...

Why we picked it

GoogleDeepMind 这次的 Gemini 4 Neon 在 Agent Arena Index 拿了 1336 分排第二,单任务成本 2.82 美元,比 Claude Opus 5.5 便宜一半还多。视频里实测了 3D 场景、游戏和 UI 生成,想做智能体应用的开发者可以看看性价比对比。

GoogleDeepMind 发布 Gemini 4 Neon,在 Agent Arena Index 上以 1336 分排名第二,成为 Google 表现最好的模型。该分数基于 4.2k 次真实智能体任务会话,超过了 #4 Carbon (High) 的 1303 分、#9 Hydrogen (High) 的 1271 分和 #11 Argon (High) 的 1229 分。在价格维度上,Neon 处于 Agent Arena 的 Pareto 前沿,每个任务的中位成本为 2.82 美元。

Image · lmarena.ai
Body

具体来看,Neon 比 #3 Claude Opus 5.5 (Max) 得分更高,单任务成本便宜 53%;比排名第一的 Claude Fable 5.2 (Max) 单任务成本便宜 69%。分项信号上,Praise vs. Complaint 达 1539 分排第 2,Steerability 1298 分排第 2,Task Completion 1316 分排第 3,说明用户对其任务结果满意度和响应修正的能力都接近头部水平。Peter Gostev 的实测视频则从 3D 场景、游戏、UI 生成和数据可视化四个方向对比了它与 Opus、Fable、Astra、Sol 的差距。

Code Arena 的 WebDev 和 Text Arena 成绩也在贴文线程中公布。对开发者而言,Neon 的意义在于以明显更低的价格拿到接近前沿的智能体任务表现,直接压缩了 Claude 系列在性价比上的优势。

Original · lmarena.ai

Just yesterday, @GoogleDeepMind released Gemini 4 Neon. Watch @petergostev’s first impressions as h...

Just yesterday, @GoogleDeepMind released Gemini 4 Neon. Watch @petergostev ’s first impressions as he unboxes this new model across 3D scenes, games, UIs and data visualizations. He compares it with Opus, Fable, Astra and Sol and shares his take on how close it is to the frontier. 00:00 Introduction and Arena ranking 00:29 Istanbul: first comparisons 01:38 White House and 3D detail 06:38 Creativity, realism and complex scenes 11:57 Live Aid and cinematic movement 13:38 Games: controls and visual quality 15:40 UI and design comparisons 17:02 Data visualization 18:03 UI coherence and design quality 21:59 Dynamic data and consistency 23:36 Pricing and model comparisons 27:30 Final verdict: how close to the frontier?

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 and AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include Gemini 4 Neon, GoogleDeepMind, Claude Opus, Agent Arena Index.

Source attribution: curated from lmarena.ai (@lmarena_ai). Cross-source corroboration: 11 independent sources have covered this development (including 宝玉, Artificial Analysis, kimmonismus), confirming industry-wide signal strength.