Google DeepMind发布Gemini 4 Neon
Google DeepMind Releases Gemini 4 Neon: 1M Context, #1 on AA Coding Agent Index
Gemini 4系列全新旗舰发布,明确切入Agent场景并在多项基准取得领先,开发者需关注其定价策略与API接入方式。
Google DeepMind just released Gemini 4 Neon. Gemini 4 Neon is a reasoning model that sits between Gemini 4 Carbon and Gemini 4 Hydrogen. It targets coding agents, knowledge work and science workflows. The developer pitch is near-frontier scores at a lower cost per task. Pricing is $3 per 1M input tokens and $15 per 1M output tokens.
Google DeepMind 刚刚发布了 Gemini 4 Neon。Gemini 4 Neon 是一个推理模型,位于 Gemini 4 Carbon 和 Gemini 4 Hydrogen 之间。它面向编码代理、知识工作和科学工作流。开发者的卖点是:在更低的单任务成本下,取得接近前沿水平的分数。定价为每 100 万输入 token 3 美元,每 100 万输出 token 15 美元。
TL;DR
TL;DR(太长不看)
- Size: 1M-token unified context, 64K default output, up to 1M via background long decode.
- Runs on: Google-hosted only. Closed weights, so no local GPU, quantized or edge builds.
- Performance: 61 on the AA Intelligence Index, ahead of Claude Opus 5.5 (59) and GPT-6.2 Sol (57).
- Best: 95.2% on GraphWalks (256K to 1M context), vs 80.1% for GPT-6.2 Sol.
- Worst: 46.2% on FrontierCode 2.0, behind Claude Opus 5.5 at 47.3%.
- Bottom line: Best: leads long-context and agentic coding at $3/$15. Worst: closed, self-reported, pricier per token than GPT-6.2 Sol.
- 规模:100 万 token 的统一上下文,默认输出 64K,通过后台长解码最高可达 1M。
- 运行环境:仅限 Google 托管。权重封闭,因此没有本地 GPU 版本、量化版或边缘端构建版。
- 性能:在 AA Intelligence Index 上得分为 61,领先于 Claude Opus 5.5(59)和 GPT-6.2 Sol(57)。
- 最佳表现:在 GraphWalks(上下文从 256K 到 1M)上准确率为 95.2%,而 GPT-6.2 Sol 为 80.1%。
- 最差表现:在 FrontierCode 2.0 上准确率为 46.2%,落后于 Claude Opus 5.5 的 47.3%。
- 总结:最佳:以 $3/$15 的价格在长上下文和代理式编码方面领先。最差:权重封闭、自我报告数据、每 token 价格高于 GPT-6.2 Sol。
What is Gemini 4 Neon?
什么是 Gemini 4 Neon?
Neon is the middle model in the Gemini 4 family. Carbon handles the hardest long-horizon projects. Hydrogen handles fast everyday execution. Neon balances depth and efficiency between the two.
Neon 是 Gemini 4 家族中的中间模型。Carbon 负责处理最困难的长期项目。Hydrogen 负责快速的日常执行。Neon 在这两者之间平衡了深度与效率。
As per the model card, Neon is adapted from Gemini 4 Carbon. Google does not disclose parameter count or architecture. Inputs cover text, images, audio and video. The knowledge cutoff is March 2026.
根据模型卡片,Neon 是基于 Gemini 4 Carbon 调整的。Google 未披露参数量或架构。输入涵盖文本、图像、音频和视频。知识截止时间为 2026 年 3 月。
How does Neon work inside agent workflows?
Neon 如何在代理工作流中运作?
Neon exposes multiple thinking levels. High is the default. xhigh is reserved for the hardest engineering, analytical and science workloads.
Neon 提供多种思考层级。High 是默认设置。xhigh 专用于最复杂的工程、分析和科学负载。
For long agent runs, Neon supports context compaction and makes fewer redundant file reads. In Google Antigravity, Neon can act as an orchestrator. It directs parallel Gemini 4 Hydrogen subagents on smaller subtasks.
对于长时间的代理运行,Neon 支持上下文压缩,并减少冗余的文件读取。在 Google Antigravity 中,Neon 可以充当协调者。它指挥并行的 Gemini 4 Hydrogen 子代理处理较小的子任务。
How does Gemini 4 Neon perform on benchmarks?
Gemini 4 Neon 在基准测试中的表现如何?
Google compares Neon against Claude Opus 5.5 and GPT-6.2 Sol. All scores below are reported by Google, not independently reproduced.
Google 将 Neon 与 Claude Opus 5.5 和 GPT-6.2 Sol 进行对比。以下所有分数均由 Google 报告,未经过独立复现验证。
- Coding: #1 on the AA Coding Agent Index at 58.6. SWE-sweep with provider harness hits 39.0%, vs 10.1% for Opus 5.5.
- Knowledge work: 56.3% on Zapier AutomationBench, vs 42.5% for Opus 5.5. 75.8% on Agents’ Last Exam v1.1.
- Science and reasoning: 84.0% on RiemannBench, 70.5% on Terminal-Bench-Science and 71.3% on HLE-Diamond.
- Long context: 95.2% on GraphWalks, vs 66.8% for Opus 5.5.
- Cyber: 90.5% on SEC-bench Pro, 70.7% on CWE-Bench v2 and 51.3% on WebExploitBench v1.2.
- 编码:在 AA Coding Agent Index 上排名第一,得分为 58.6。使用 provider harness 的 SWE-sweep 命中率为 39.0%,而 Opus 5.5 仅为 10.1%。
- 知识工作:在 Zapier AutomationBench 上得分为 56.3%,而 Opus 5.5 为 42.5%。在 Agents’ Last Exam v1.1 上得分为 75.8%。
- 科学与推理:在 RiemannBench 上得分为 84.0%,在 Terminal-Bench-Science 上得分为 70.5%,在 HLE-Diamond 上得分为 71.3%。
- 长上下文:在 GraphWalks 上准确率为 95.2%,而 Opus 5.5 为 66.8%。
- 网络安全:在 SEC-bench Pro 上准确率为 90.5%,在 CWE-Bench v2 上为 70.7%,在 WebExploitBench v1.2 上为 51.3%。
Neon does not win every row. Opus 5.5 leads Terminal-bench 4.1 (65.8% vs 65.2%) and FrontierCode 2.0 (47.3% vs 46.2%). GPT-6.2 Sol leads Chartography (75.1% vs 73.5%).
Neon 并非在所有行中都获胜。Opus 5.5 在 Terminal-bench 4.1(65.8% vs 65.2%)和 FrontierCode 2.0(47.3% vs 46.2%)上领先。GPT-6.2 Sol 在 Chartography 上领先(75.1% vs 73.5%)。
On the Vals Index, Neon scores 70.9% and ranks #2. On the Agent Arena index it also ranks #2, with a score of 1336.
在 Vals Index 上,Neon 得分为 70.9%,排名第 2。在 Agent Arena 指数上,它也排名第 2,得分为 1336。
What does Gemini 4 Neon cost?
Gemini 4 Neon 的成本是多少?
Neon lists at $3 input, $0.15 cached input and $15 output per 1M tokens. Google says Neon at high thinking costs 32% less per task than Carbon. It also claims 50% to 80% lower cost per task vs rivals on Vals Index and Agent Arena. At xhigh, Google cites 36%+ lower cost than rivals at their highest settings.
Neon 的定价为每百万 token 输入 3 美元、缓存输入 0.15 美元、输出 15 美元。Google 表示,在高思考模式下,Neon 每项任务的成本比 Carbon 低 32%。它还声称,在 Vals Index 和 Agent Arena 上,其每项任务的成本比竞争对手低 50% 到 80%。在 xhigh 模式下,Google 引用数据称,其成本比竞争对手在其最高设置下的成本低 36% 以上。
Those are per-task claims, not per-token prices. GPT-6.2 Sol lists at $2/$10. Claude Opus 5.5 lists at $4/$20. Gemini 4 Carbon is $2/$10 until November 30, then $4/$20.
这些是每项任务的声明,而非每个 token 的价格。GPT-6.2 Sol 的定价为 2 美元/10 美元。Claude Opus 5.5 的定价为 4 美元/20 美元。Gemini 4 Carbon 在 11 月 30 日之前为 2 美元/10 美元,之后为 4 美元/20 美元。
Where can developers use Gemini 4 Neon?
开发者可以在哪里使用 Gemini 4 Neon?
- Gemini API paid tier, via Google AI Studio and Android Studio
- Gemini for business, via API and the model dropdown
- Google Antigravity, for paid Google AI Pro and Ultra subscribers
- The Gemini app, for Pro and Ultra subscribers
- Gemini API 付费层级,通过 Google AI Studio 和 Android Studio
- 面向企业的 Gemini,通过 API 和模型下拉菜单
- Google Antigravity,面向付费的 Google AI Pro 和 Ultra 订阅用户
- Gemini 应用,面向 Pro 和 Ultra 订阅用户
What do the safety numbers show?
安全数据显示了什么?
Neon runs on the Gemini 4 safety and security stack. Google reports a 0.09% misaligned action rate on its agentic safety suite. It passes 99.2% of the agent honesty evaluation. It shows a 0.0% attack success rate on Gray Swan’s prompt injection benchmark at the highest thinking level. Its 2.45% shortcut rate on a coding cheating test is the lowest of any Gemini 4 model.
Neon 运行在 Gemini 4 的安全与防护堆栈之上。Google 报告称,在其代理安全套件中,未对齐行为率为 0.09%。它通过了 99.2% 的代理诚实度评估。在 Gray Swan 的提示注入基准测试的最高思考级别下,其攻击成功率为 0.0%。在编码作弊测试中,其 2.45% 的捷径率是任何 Gemini 4 模型中最低的。
Gemini 4 Neon vs closest competitors
Gemini 4 Neon 与最接近的竞争对手对比
| Feature | Gemini 4 Neon | Gemini 4 Carbon | Claude Opus 5.5 | GPT-6.2 Sol |
|---|---|---|---|---|
| Developer | Google DeepMind | Google DeepMind | Anthropic | OpenAI |
| Parameters | Not disclosed | Not disclosed | Not disclosed | Not disclosed |
| Weights | Closed | Closed | Closed | Closed |
| Context window | 1M | Not disclosed in launch post | 1M | 1M |
| Max output | 64K default, 1M background | Not disclosed | 128K | Not disclosed |
| Input / output per 1M | $3 / $15 | $2 / $10 to Nov 30, then $4 / $20 | $4 / $20 | $2 / $10 |
| AA Intelligence Index | 61 | Not disclosed | 59 | 57 |
| Vals Index v2.2* | 70.9% | Not disclosed | 70.1% | 66.1% |
| Terminal-bench 4.1* | 65.2% | Not disclosed | 65.8% | 61.5% |
| SWE-sweep (provider harness)* | 39.0% | Not disclosed | 10.1% | 8.3% |
| GraphWalks 256K to 1M* | 95.2% | Not disclosed | 66.8% | 80.1% |
| 功能 | Gemini 4 Neon | Gemini 4 Carbon | Claude Opus 5.5 | GPT-6.2 Sol |
|---|---|---|---|---|
| 开发者 | Google DeepMind | Google DeepMind | Anthropic | OpenAI |
| 参数量 | 未披露 | 未披露 | 未披露 | 未披露 |
| 权重 | 闭源 | 闭源 | 闭源 | 闭源 |
| 上下文窗口 | 1M | 发布帖中未披露 | 1M | 1M |
| 最大输出 | 默认 64K,后台 1M | 未披露 | 128K | 未披露 |
| 每百万 token 输入/输出价格 | $3 / $15 | 11月30日前 $2 / $10,之后 $4 / $20 | $4 / $20 | $2 / $10 |
| AA Intelligence Index | 61 | 未披露 | 59 | 57 |
| Vals Index v2.2* | 70.9% | 未披露 | 70.1% | 66.1% |
| Terminal-bench 4.1* | 65.2% | 未披露 | 65.8% | 61.5% |
| SWE-sweep (provider harness)* | 39.0% | 未披露 | 10.1% | 8.3% |
| GraphWalks 256K to 1M* | 95.2% | 未披露 | 66.8% | 80.1% |
*Scores from Google’s launch comparison table, self-reported. AA scores for Opus 5.5 and GPT-6.2 Sol match Artificial Analysis.
*分数来自 Google 的发布对比表,由 Google 自行报告。Opus 5.5 和 GPT-6.2 Sol 的 AA 分数与 Artificial Analysis 一致。
Key Takeaways
关键要点
- Neon sits between Carbon and Hydrogen, priced at $3/$15 per 1M tokens.
- Scores 61 on AA Intelligence Index, #1 on AA Coding Agent Index.
- Biggest lead: 95.2% on GraphWalks long-context recall.
- Trails Opus 5.5 on Terminal-bench 4.1 and FrontierCode 2.0.
- Closed and API-only, with Google-reported benchmarks.
- Neon 位于 Carbon 和 Hydrogen 之间,每百万 token 定价为 3 美元/15 美元。
- 在 AA 智能指数中得分 61,在 AA 编码代理指数中排名第一。
- 最大优势:在 GraphWalks 长上下文召回率上达到 95.2%。
- 在 Terminal-bench 4.1 和 FrontierCode 2.0 上落后于 Opus 5.5。
- 仅限闭源和 API 调用,采用 Google 报告的基准测试数据。
Check out the launch post and the model card. You can check the deployment options and specs here through our GPULeaderboard.com chrome extension.
查看发布文章和模型卡片。您可以通过我们的 GPULeaderboard.com Chrome 扩展程序在此处查看部署选项和规格。
Also, feel free to follow us on Twitter and don’t forget to join our 150k+ML SubReddit and Subscribe to our Newsletter. Wait! are you on telegram? now you can join us on telegram as well.
此外,欢迎在 Twitter 上关注我们,别忘了加入我们有 15 万+成员的 ML SubReddit 并订阅我们的新闻通讯。等等!你在 Telegram 上吗?现在你也可以在 Telegram 上加入我们。
更进一步:量化金融体系
看懂新闻只是起点——沿量化金融路径,把它变成能交付的工程能力